CN109103929B - Power distribution network economic optimization scheduling method based on improved dynamic kriging model - Google Patents

Power distribution network economic optimization scheduling method based on improved dynamic kriging model Download PDF

Info

Publication number
CN109103929B
CN109103929B CN201811057761.5A CN201811057761A CN109103929B CN 109103929 B CN109103929 B CN 109103929B CN 201811057761 A CN201811057761 A CN 201811057761A CN 109103929 B CN109103929 B CN 109103929B
Authority
CN
China
Prior art keywords
power
wind
output
model
hour
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201811057761.5A
Other languages
Chinese (zh)
Other versions
CN109103929A (en
Inventor
马丽叶
王志强
谷林
万福瑞
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Wuxi Xiangyuan Information Technology Co ltd
Original Assignee
Yanshan University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Yanshan University filed Critical Yanshan University
Priority to CN201811057761.5A priority Critical patent/CN109103929B/en
Publication of CN109103929A publication Critical patent/CN109103929A/en
Application granted granted Critical
Publication of CN109103929B publication Critical patent/CN109103929B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • H02J3/383—
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
    • G06Q10/06312—Adjustment or analysis of established resource schedule, e.g. resource or task levelling, or dynamic rescheduling
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
    • G06Q50/06—Electricity, gas or water supply
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J15/00—Systems for storing electric energy
    • H02J15/003—Systems for storing electric energy in the form of hydraulic energy
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/24—Arrangements for preventing or reducing oscillations of power in networks
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/28—Arrangements for balancing of the load in a network by storage of energy
    • H02J3/386—
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J3/00—Circuit arrangements for ac mains or ac distribution networks
    • H02J3/38—Arrangements for parallely feeding a single network by two or more generators, converters or transformers
    • H02J3/46—Controlling of the sharing of output between the generators, converters, or transformers
    • H—ELECTRICITY
    • H02—GENERATION; CONVERSION OR DISTRIBUTION OF ELECTRIC POWER
    • H02J—CIRCUIT ARRANGEMENTS OR SYSTEMS FOR SUPPLYING OR DISTRIBUTING ELECTRIC POWER; SYSTEMS FOR STORING ELECTRIC ENERGY
    • H02J2203/00—Indexing scheme relating to details of circuit arrangements for AC mains or AC distribution networks
    • H02J2203/20—Simulating, e g planning, reliability check, modelling or computer assisted design [CAD]
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E10/00—Energy generation through renewable energy sources
    • Y02E10/50—Photovoltaic [PV] energy
    • Y02E10/56—Power conversion systems, e.g. maximum power point trackers
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E10/00—Energy generation through renewable energy sources
    • Y02E10/70—Wind energy
    • Y02E10/76—Power conversion electric or electronic aspects
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
    • Y02E70/00—Other energy conversion or management systems reducing GHG emissions
    • Y02E70/30—Systems combining energy storage with energy generation of non-fossil origin

Abstract

The invention discloses a power distribution network economic optimization scheduling method based on an improved dynamic kriging metal model, which comprises the following contents: modeling the state of the pumped storage unit; modeling output of a wind turbine generator and user load; modeling a photovoltaic generator set; providing relevant constraint conditions of combined output constraint of the wind-light-water combined system, and modeling the wind-light-water combined system; modeling the economic optimization operation of the power distribution network by using the constraint conditions in the step 4; randomly extracting sample points by using Monte-Carlo simulation to obtain corresponding system variables and construct an initial sample library; constructing a gradient enhanced dynamic Kriging model, adding gradient information, and screening to obtain a new sample library; and solving the economic optimization model of the power distribution network by using a BCC optimization algorithm on the basis of the new sample library. The invention reduces the power fluctuation influence caused by the grid connection of the new wind and light energy sources and smoothes the wind power and the photovoltaic power generation power, improves the power generation stability of the new wind and light energy sources, increases the consumption of the new wind and light energy sources and reduces the investment and operation cost of a power distribution network system.

Description

Power distribution network economic optimization scheduling method based on improved dynamic kriging model
Technical Field
The invention relates to the field of new energy power generation operation, in particular to a power distribution network economic optimization scheduling method based on an improved dynamic kriging metal model.
Background
Wind energy and solar energy are widely regarded as renewable green energy, along with the rapid development of wind power generation technology and solar photovoltaic power generation technology, development of new energy has a potential of being unblocked, and wind power generation and solar photovoltaic power generation have the characteristics of inaccuracy in prediction, randomness and instability, so that wind power generation and photovoltaic power generation cannot be effectively utilized. Not only can a large amount of wind energy become abandoned wind, but also the consumption of the wind power is greatly reduced, and the new energy power generation equipment is bought at high cost to cause the wind power not to be laid out. And the access of large-scale wind power and photovoltaic power generation equipment can greatly increase the instability of the output power of the power distribution network system, and influence the aspects of optimized scheduling, system stability, electric energy quality and the like in the operation process of the power system. Aiming at the problems, the power output is stabilized in a reasonable mode, the air abandoning amount is reduced, new energy is fully utilized, and the low-carbon economic dispatching of the power system has important social significance.
In recent years, with the continuous development of new energy, energy storage devices have become key accessories of wind power generation systems. The energy storage device comprises a storage battery, an air compression energy storage device and the like, wherein the storage battery is mostly selected as an energy storage device, however, the storage battery is used as a mode for storing electric energy, the storage battery has the advantages that the release time can be from hours to days, the energy conversion rate is high (70% -85%), the storage battery can indirectly store electric energy, when wind energy is excessive, redundant wind energy is used for driving a water pump, water is pumped from a lower reservoir to an upper reservoir to be stored, then the water is discharged to generate electricity when the wind energy is insufficient, and the electricity flows into the lower reservoir.
At the present stage, for the problem of wind power-photovoltaic power generation and pumped storage combined optimization operation, experts and scholars at home and abroad carry out some exploratory work, and most of the research focuses on processing the problems of wind power-photovoltaic power generation and pumped storage combined operation strategy, scheduling model optimization solution, capacity optimization configuration of each unit in the system and the like, but the optimization operation cannot be directly related to the system investment cost.
In summary, it is necessary to invent a new power distribution network optimization scheduling research strategy, which not only increases the consumption of wind power generation and photovoltaic power generation, but also increases the stability of thermal power generating unit output, and simultaneously minimizes the total investment cost of the power distribution network system.
Disclosure of Invention
The invention aims to provide a power distribution network economic optimization scheduling method based on an improved dynamic kriging model, and the method has the characteristics of high efficiency, comprehensive consideration factors and strong practicability.
In order to realize the purpose, the invention is realized by the following technical scheme:
a power distribution network economic optimization scheduling method based on an improved dynamic kriging model comprises the following steps:
step 1, modeling the state of a pumped storage unit;
step 2, modeling output of the wind turbine generator and user load;
step 3, modeling the photovoltaic generator set;
step 4, providing relevant constraint conditions of combined output constraint of the wind-light-water combined system, and modeling the wind-light-water combined system;
step 5, modeling the economic optimization operation of the power distribution network by using the constraint conditions in the step 4;
step 6, randomly extracting sample points by using Monte-Carlo simulation to obtain corresponding system variables, and constructing an initial sample library;
step 7, constructing a gradient enhanced dynamic Kriging model, and adding gradient information to screen an initial sample library;
and 8, solving the economic optimization model of the power distribution network by using a BCC optimization algorithm on the basis of the new sample library.
Further, in step 1, the pumped storage group state is modeled, and the specific process is as follows:
the pumped storage unit, the wind turbine generator and the photovoltaic generator are combined to form a new combined unit, so that the advantages of the pumped storage unit can be fully exerted, and the output of the wind turbine generator and the photovoltaic generator is effectively smoothed.
The method of the invention determines the running state of the pumped storage unit according to the predicted value of the wind power output at the day before, introduces 0-1 integer variable to consider two stages of pumping and generating of the unit, and establishes a state model of the pumped storage unit as follows:
X(t,k)+Y(t,k)＝1 (1)
(1) in the formula: when X (t, k) is 1, the unit is a power generation stage in the kth period of the t hour, the coordination period is 24 hours, four periods are divided every hour, and k is 1,2,3, 4; when Y (t, k) is 1, the unit is a water pumping stage.
Further, in step 2, modeling the wind turbine generator output and the user load is performed in the following specific process:
2.1 modeling the output of the wind turbine
1) Output characteristic of single wind turbine
The wind speed is a random variable complying with Weibull distribution, and a power characteristic curve of the wind turbine generator is generally provided by a fan manufacturer and can also be obtained through actual measurement; in the calculation, according to the power characteristic curve of the wind turbine generator, the output power P of a single wind turbine generatorIFThe relationship v to wind speed can be approximated by a piecewise function of:
Figure GDA0003012943650000041
(2) in the formula: v. ofrAnd PWrRespectively representing the rated wind speed and the rated power of the fan in the kth time period of the tth hour; v. ofinTo cut into the wind speed; v. ofoutTo cut out windSpeed;
2) describing characteristics of total output curve of wind turbine generator
Total generated power P of wind turbineWCan be expressed as:
Pw(t,k)＝efNW(t,k)PIF(t,k) (3)
0≤PIF(t,k)≤PWR (4)
(3) formulae (I) and (4): pIF(t,k),Pw(t, k) respectively representing the generating power of a single fan and the total wind turbine generator in the kth period of the tth hour, wherein the unit is MW; e and f are respectively the transmission efficiency and the power generation efficiency of the fan, and are expressed in percent; n is a radical ofW(t, k) is the number of fans which normally operate in the wind power plant in the kth time period in the tth hour; pWR(t, k) is rated power of each unit, and the unit is MW;
2.2 user load modeling
The method adopts a simple exponential smoothing model and an ARIMA model in a time sequence prediction module to model the user load, tries to establish a mathematical model according to the historical data of the load, and then establishes a mathematical expression of load prediction on the basis of the mathematical model to predict the future load;
simple exponential smoothing model
The simple model is one of exponential smoothing models, which is a nonlinear estimation method, and the basic principle is to minimize the Mean Square Error (MSE) between the predicted value and the observed value; the method is applicable to time series data that is free of trending and seasonal components;
the basic prediction formula is:
St＝Ayt+(1-A)St-1 (5)
the general predictive formula is:
St＝Ayt+(1-A)yt-1+…+(1-A)t-2y2+(1-A)t-1y1 (6)
(5) formulae (I) and (6): y istIs the observed data at time t; stThe smoothed data; a is oneA real number between 0 and 1;
the ARIMA Model (Autoregressive Integrated Moving Average Model, ARIMA) is a most applied time series prediction Model, and can analyze time series data containing seasonal components, and comprises 3 main parameters: the method comprises the following steps of (1) carrying out autoregressive order p, differential order d and moving average order q, wherein the general model is recorded as ARIMA (p, d and q);
difference
When the model is used, firstly, the time series is stabilized through difference, and the difference is divided into general difference and seasonal difference;
the general difference formula is:
Figure GDA0003012943650000051
wherein y istIs the original time sequence, B is the delay operator,
Figure GDA0003012943650000052
in order to be a first-order difference,
Figure GDA0003012943650000053
is a d-order difference; the seasonal difference formula is:
Figure GDA0003012943650000061
wherein: y istIs a sequence with a period of T,
Figure GDA0003012943650000062
representing a seasonal difference operator;
② autoregressive moving average model ARMA (p, q)
Xt＝φ1Xt-1+φ2Xt-2+…φpXt-p+θ1εt-1+θ2εt-2+…+θqεt-q+εt (9)
Wherein epsilontFor white noise sequence, equation (9) represents time sequence XtThe autocorrelation function and the partial autocorrelation function of the model are respectively nonzero after the order of p and q, namely, the model has trailing property.
Further, in step 3, modeling the photovoltaic generator set, specifically solving the following process:
the generated power of a photovoltaic power generation module depends on three parameters: solar irradiance, field ambient temperature, and the characteristics of the module itself; solar radiation is modeled in the manner described below:
Figure GDA0003012943650000063
(10) in the formula: s is solar irradiance (kw/m)2) α and β are parameters of the beta probability distribution function;
Ppv(s)＝N×FF×V(s)×I(s) (11)
Figure GDA0003012943650000064
V(s)＝Voc-Kv×Tc (13)
I(s)＝Sa×[Isc+Ki(Tc-25)] (14)
Figure GDA0003012943650000065
(11) [ formula (15) ]: t iscIs the cell temperature, T, in degrees CelsiusAIs the ambient temperature in degrees Celsius, KvAnd KiThe temperature coefficients of voltage and current are V/DEG C, A/DEG C and NOTIs the nominal operating temperature of the photovoltaic cell, in degrees C, FF is the charge factor, IscIs an electric currentShort-circuit current in the characteristic, VocIs an open circuit voltage in the voltage characteristic, IMPPAnd VMPPIs the current and voltage at the maximum power point in the current and voltage characteristics; saIs the average solar irradiance; ppv(s) represents photovoltaic power generation power per unit solar irradiance;
photovoltaic power generation P in the method of the inventionpvAnd (t, k) is obtained from the measured data and is expressed as the generated power of the photovoltaic cell generator set under the condition of fully utilizing all solar irradiance in the kth hour.
Further, in step 4, the wind, light and water combined system is modeled by proposing relevant constraint conditions of combined output constraint of the wind, light and water combined system, and the concrete implementation process is as follows:
4.1 wind-solar-water combined system modeling
Considering that wind power generation and solar photovoltaic power generation have the characteristics of inaccurate prediction, randomness and instability, the wind power generation, the photovoltaic power generation and pumped storage form a wind-light-water combined power generation system, and the wind-light-water combined system has the following specific coordination mode: because the coordination time period of the wind-light storage combined system in the day-ahead scheduling can be effectively shortened after the pumped storage unit is added, the method takes 24 hours a day as a cycle, and is divided into four time periods per hour, and 96 coordination time periods are totally included; when the sum of the output of the wind generating set and the output of the photovoltaic generating set in a certain hour is larger than or equal to the average value of the output in the period, the hour is defined as a water pumping state, the wind and light generating capacity provided to the system in each time period in the hour is the same by controlling the water pumping power in the stage, and redundant wind and light power generation is used for pumping the water storage water pumping output; when the sum of the output of the wind generating set and the output of the photovoltaic cell generating set in a certain hour is smaller than the average value of the predicted output of wind power, the hour is defined as a power generation state, the wind-solar power generation at the stage is less, at the moment, water is discharged from a reservoir for power generation, and the power generation power is controlled to ensure that the electric quantity generated by the water pumping set in each period of the hour is the same as the electric energy provided to the system after the coordination of the electric quantity generated by the wind-solar power generation; therefore, the combined output value of the combined system in each pumping state and power generation state is obtained, and the wind-light energy storage combined power generation system is arrangedP for combined power generationws(t, k) represents a joint force output value at the kth time at the tth hour in one cycle;
4.2 associated constraint of combined output constraint of wind, light and water combined system
1) The wind power storage combined system ensures that the coordinated total output of each time period in an hour is kept stable, and the output active power of the wind power plant is relatively smooth:
Figure GDA0003012943650000071
Pws(t,k)＝Pws(t,k+1) (17)
wherein:
Pws(t,k)＝Pt XX(t,k)+Pt YY(t,k) (18)
Pp(t,k)＝(Pw(t,k)+Ppv(t,k))Y(t,k)-Pt Y (19)
Pg(t,k)＝Pt X-(Pw(t,k)+Ppv(t,k))X(t,k) (20)
X(t,k)+Y(t,k)＝1 (21)
Figure GDA0003012943650000081
in the formula: pws(t, k) is the wind-solar-storage combined power generation power in the kth time period of the tth hour; pt Y,Pt XThe wind-solar energy storage combined system is used for generating combined power in a water pumping state and a power generation state respectively; pw(t, k) is the generating power of the wind turbine generator in the kth time period of the tth hour; ppv(t, k) is the power generated by the photovoltaic generator set in the kth time period of the tth hour; pp(t, k) and Pg(t, k) respectively representing the pumping power and the generating power of the pumped storage unit in the kth time period of the tth hour; pwp.tThe sum of the generated power of the wind turbine generator set and the photovoltaic generator set in the tth hour; t is a period, namely 24 hours; k isThe number of time periods in each hour is 4 time periods;
2) wind-solar-storage combined output constraint:
Pmin(t,k)≤Pws(t,k)≤Pmax(t,k) (23)
(23) in the formula: pmin(t,k)、Pmax(t, k) is the minimum value and the maximum value of power transmitted to the power grid by the wind-solar-storage combined system in the kth period in the tth hour;
3) the wind-solar storage combined system is restricted in pumping power and generating power:
Figure GDA0003012943650000082
Figure GDA0003012943650000083
0≤PP(t,k)≤Ppmax(t,k) (26)
in the formula: etagGenerating efficiency for the water pump; ppmax(t, k) and Pgmax(t, k) are respectively the upper limit of the pumping power and the upper limit of the generating power; e (t, k) is the storage capacity of the pumped storage power station; the interval between two adjacent time intervals is delta T;
4) wind power-actual generated power constraint of photovoltaic generator set:
Pw.min(t,k)≤Pw(t,k)≤Pw.max(t,k) (27)
(27) in the formula: pw.min(t, k) and Pw.max(t, k) is the minimum value and the maximum value of the installed capacity of the wind turbine generator in the kth time period of the tth hour;
PPv.min(t,k)≤Ppv(t,k)≤PPv.max(t,k) (28)
(28) in the formula: ppv.min(t, k) and Ppv.maxAnd (t, k) is the minimum value and the maximum value of the generated power of the photovoltaic generator set in the kth period of the tth hour.
Further, in step 5, the economic optimization operation of the power distribution network is modeled by using the constraint conditions in step 4, and the specific implementation process is as follows:
generally, wind power and photovoltaic power should not be wasted according to national energy policy, i.e. incorporating the requirements of all clean energy. And as the wind and light renewable energy sources which are preferentially scheduled are merged into the power grid, the output intermittence and the fluctuation of the thermal power generating unit are poorer and poorer, so that the cost of the whole power distribution network system is increased. In order to reduce the consumption of non-renewable energy sources and the output intermittence and fluctuation of a thermal power generating unit, the renewable energy sources are maximally consumed, and the system cost of a power distribution network is reduced as much as possible; output power P of single wind turbineIFThe approximate piecewise function of the relation v with the wind speed can be known, the invention provides that the total output of the wind-light storage combined system in the period is maximum on the basis of meeting the requirement that the combined output of the wind-light storage combined system changes along with the change of the load requirement, and the minimum output and the more stable output of the thermal power generating unit can be realized; according to experience, the investment cost of the thermal power generating unit occupies a large proportion, so that the cost of the whole power distribution system can be reduced;
1) objective function
The optimization operation model objective function of the power distribution network consists of two parts, namely construction cost and energy consumption cost, and the concrete model is as follows:
Min:Ctotal＝CM+CQ+CC+CR (29)
(29) in the formula: ctotalFor the total cost of the distribution network system, CMFor the cost of coal for thermal power generating units, CQFor exhaust gas emission costs, CCFor operating and maintaining the equipment in the power supply network system, CRThe construction cost of the whole life cycle of the power distribution network system is balanced to the cost of each year;
2) optimizing constraints
In order to ensure the stable operation of the power distribution network, the system needs to meet the demand and supply balance;
the system needs to satisfy the following energy balance equation:
PG(t,k)+Pws(t,k)＝yL(t,k) (30)
(30) in the formula: y isL(t, k) is the user negative at the kth moment of the tth hourThe load demand electric quantity is a continuous and derivable function of the user load in a period; pws(t, k) is the power generation power of the wind-light-water combined system at the kth moment in the tth hour, and is a continuous derivative function in a period; pGAnd (t, k) is the power generated by the thermal power generating unit at the kth moment in the tth hour.
Further, in step 6, the monte-talo simulation is used to randomly extract sample points, obtain corresponding system variables, and construct an initial sample library, which is specifically implemented as follows:
step 5, knowing that the power distribution network consists of a wind-solar-energy storage combined system and a thermal power generating unit; on the premise of preferentially scheduling renewable energy sources, and an optimized output scheme of the wind-light-storage combined unit can be obtained in the wind-light-water combined system model established in the step 4, so that the method selects the optimized output P of the wind-light-storage combined unit at the kth time of the tth hour in a periodws(t, k) are random variables, and the value range of the variables meets the wind-solar-storage combined output constraint Pmin(t,k)≤Pws(t,k)≤Pmax(t,k)；
The method adopts the random simulation technology to randomly extract sampling points for a coordination time interval every 15 minutes, so that the extracted sample points can be uniformly distributed in the whole sampling space at equal probability; the obtained initial sample library also provides important basis and guarantee for the construction and the correction of a subsequent gradient enhanced Kriging model;
selecting the total generated power P in the periodws.nAs system variables:
Figure GDA0003012943650000111
the plane formed by the output of the wind-solar-energy-storage combined unit at each moment in a period is uniformly divided into 96 subintervals, n sample points are randomly extracted in each subinterval by utilizing a random simulation technology, 96 multiplied by n random variables are collected to form an initial sample library Z, and the expression is as follows:
Figure GDA0003012943650000112
carrying out calculation analysis on 96 multiplied by n initial samples selected randomly to obtain the total generated power P in n periodsws.nAs a function of the system variables,
S＝[X1,…Xn] (33)
compiling a MINLP solving program under a Matlab platform by using the unit models, the energy balance equations and the unit output constraint conditions and combining a BCC (bacterial colony chemotaxis algorithm) optimization algorithm to solve the economic optimization operation model of the power distribution network, so that the most economic total cost of the power distribution network under n corresponding design schemes can be obtained as corresponding function values;
YS＝[Y1,…Yn]＝[Y1(X1),…Yn(Xn)] (34)
according to the analysis, the total generating power in the period of the wind-light-storage combined unit and the optimized total system cost are respectively used as coordinate values to form a two-dimensional initial particle library Q which is recorded as (N)1,N2,…Nn) And the coordinates of the nth particle are recorded as Nn(Xn,Yn) Wherein X isnRepresenting the wind-solar total generated power P of the nth sample particlews.n，YnRepresents the most economic total cost of the distribution network under the configuration scheme corresponding to the nth sample particle, NnThe system composition scheme representing the power distribution network can be recorded as scheme Nn。
Further, in step 7, the step of constructing a gradient enhanced dynamic Kriging model, and adding gradient information to screen an initial sample library includes the following specific implementation processes:
in order to meet the requirement that the value of the system variable proposed in the step 5 is maximum on the basis of changing along with the change of the user load requirement, the method improves the precision of the kriging model by introducing gradient information, and evolves to be a new agent model method, namely a gradient enhanced kriging model; the accuracy of the proxy model is improved by introducing first-order partial derivative information; the specific analysis is as follows:
in order to screen an initial sample library, n × m design variables and corresponding n × m partial derivative values under a wind-light-storage combined output curve are selected near random variables; the information of the sample library after screening and the partial derivative thereof is as follows:
Figure GDA0003012943650000121
Figure GDA0003012943650000122
Figure GDA0003012943650000123
in the gradient enhanced Kriging model constructed by the screened sample library, because the number of sample points is too large, gradient information is required to be added to screen characteristic sample points to optimize the model, so that the characteristic sample points meeting the gradient information need to be screened according to constraint conditions to modify the Kriging model so as to improve the accuracy of the model; obtaining the optimal sample library S after screeningNReuse the sample library SNOptimizing a gradient enhanced dynamic Kriging model to finally obtain a relatively accurate global optimal solution;
the selection of the characteristic sample points in the method is mainly based on the following gradient information:
the random variable is changed along with the change of the load requirement at each sampling position; the partial derivative change formula of the load in each sampling interval is as follows:
Figure GDA0003012943650000124
screening partial derivative information of each column of the formula (37) by using a formula (38), removing sample information which does not meet the conditions, and constructing an optimal sample library SN：
SN＝(X1,X2…XN)。 (39)
Further, in step 8, the BCC optimization algorithm is used to solve the economic optimization model of the power distribution network based on the new sample library, and the specific implementation process is as follows:
by utilizing the unit models, the energy balance equations and the unit output constraint conditions mentioned in the step 4 and the new sample library in the step 7, the wind-light-storage combined output scheme under N corresponding design schemes can be obtained, a MINLP program is compiled and solved under a Matlab platform by combining a BCC (bacterial population chemotaxis algorithm) optimization algorithm, and the optimal economic energy utilization scheme and the system operating cost of the power distribution network can be obtained through optimization simulation calculation; storing and establishing optimal particle library
Figure GDA0003012943650000131
And comparing with the constructed initial particle library in the step 6 to verify the effectiveness of the strategy provided by the invention.
Due to the adoption of the technical scheme, compared with the prior art, the invention has the following beneficial effects:
compared with the prior art, the economic dispatching method of the power distribution network based on the improved dynamic kriging model has the characteristics of high efficiency, comprehensive consideration factors and strong practicability, can effectively reduce the power fluctuation influence caused by wind and light new energy grid connection and smooth wind power and photovoltaic power generation power, improves the power generation stability of wind and light new energy, increases the consumption of the wind and light new energy, and reduces the investment and operation cost of the power distribution network system.
The invention fully considers the inaccurate prediction, randomness and instability of wind power generation and solar photovoltaic power generation, not only can smooth the power output of wind power generation and photovoltaic power generation by configuring a pumped storage system, increases the stability of the power generation power of a thermal power generating unit, solves the problem of power fluctuation caused by the fact that new wind and photovoltaic energy is merged into a power grid, increases the consumption of wind power generation and photovoltaic power generation, reduces the consumption of non-renewable energy, and finally achieves the purpose of minimizing the operation cost of a power distribution network. A simple, convenient, direct and effective strategy is provided for economic optimization operation of the power distribution network.
Drawings
FIG. 1 is a block diagram of the method system of the present invention;
FIG. 2 is a diagram of the coordination mode of wind power-photovoltaic power generation and pumped storage in the method of the present invention;
FIG. 3 is a general flow diagram of the method of the present invention.
Detailed Description
The invention is described in further detail below with reference to the following detailed description and accompanying drawings:
the invention discloses a power distribution network economic optimization scheduling method based on an improved dynamic kriging model, which is provided under the condition that a large-scale distributed power supply is considered to be connected into an active power distribution network.
With reference to fig. 1,2 and 3, the method of the present invention comprises the following steps:
1. defining the running state of the pumped storage unit, and establishing a state model of the pumped storage unit;
the pumped storage unit, the wind turbine generator and the photovoltaic cell generator are combined to form a new combined unit, so that the advantages of the pumped storage unit can be fully exerted, and the output of the wind turbine generator and the photovoltaic cell generator is effectively smoothed. The method of the invention determines the running state of the pumped storage unit according to the predicted value of the wind power output at the day before, introduces 0-1 integer variable to consider two stages of pumping and generating of the unit, and establishes a state model of the pumped storage unit as follows:
X(t,k)+Y(t,k)＝1 (40)
(40) in the formula: when X (t, k) is 1, it means that the unit is in the kth (the coordination period is 24 hours, and each hour is divided into four time intervals, where k is 1,2,3,4) time interval is a power generation phase, and when Y (t, k) is 1, the unit is a water pumping phase.
2. Modeling output of a wind turbine generator and user load;
1) output characteristic of single wind turbine
The wind speed is a random variable complying with Weibull distribution, and a power characteristic curve of the wind turbine generator is generally provided by a wind turbine manufacturer and can also be obtained through actual measurement. In the calculation, according to the power characteristic curve of the wind turbine generator, the output power P of a single wind turbine generatorIFThe relationship v to wind speed can be approximated by a piecewise function of:
Figure GDA0003012943650000151
(41) in the formula: v. ofrAnd PWrRespectively representing the rated wind speed and the rated power of the fan in the kth time period of the tth hour; v. ofinTo cut into the wind speed; v. ofoutCutting out the wind speed;
2) describing characteristics of total output curve of wind turbine generator
The total generated power PW of the wind turbine may be expressed as:
Pw(t,k)＝efNW(t,k)PIF(t,k) (42)
0≤PIF(t,k)≤PWR (43)
(42) and (43) in the formulae: pIF(t,k),Pw(t, k) which are respectively the generating power (MW) of a single fan and the total wind turbine generator in the kth period of the tth hour; e and f are respectively the transmission efficiency and the power generation efficiency (%) of the fan; n is a radical ofW(t, k) is the number of fans which normally operate in the wind power plant in the kth time period in the tth hour; pWRAnd (t, k) is the rated power (MW) of each unit.
The method adopts a simple exponential smoothing model and an ARIMA model in a time sequence prediction module to model the user load, tries to establish a mathematical model according to the historical data of the load, and then establishes a mathematical expression of load prediction on the basis of the mathematical model to predict the future load.
Simple exponential smoothing model
The simple model is one of exponential smoothing models, which is a non-linear estimation method, and the basic principle is to minimize the Mean Square Error (MSE) between the predicted value and the observed value. The method is applicable to time series data that does not contain trends and seasonal components.
Basic predictive formula:
St＝Ayt+(1-A)St-1 (44)
general predictive formula:
St＝Ayt+(1-A)yt-1+…+(1-A)t-2y2+(1-A)t-1y1 (45)
(44) and (45) in the formula: y istIs the observed data at time t; stThe smoothed data; a is a real number between 0 and 1.
The ARIMA model is a most widely used time series prediction model, and can analyze time series data containing seasonal components, and comprises 3 main parameters: the autoregressive order (p), the difference order (d) and the moving average order (q), and the general model is denoted as ARIMA (p, d, q).
Difference
When using this model, the time series is first smoothed by differences, which are classified into general differences and seasonal differences.
General difference formula:
Figure GDA0003012943650000161
wherein y istIs the original time sequence, B is the delay operator,
Figure GDA0003012943650000162
in order to be a first-order difference,
Figure GDA0003012943650000163
is the d-order difference.
Seasonal difference formula:
Figure GDA0003012943650000164
wherein Y istIs a sequence with a period of T,
Figure GDA0003012943650000165
representing a seasonal difference operator.
② autoregressive moving average model ARMA (p, q)
Xt＝φ1Xt-1+φ2Xt-2+…φpXt-p+θ1εt-1+θ2εt-2+…+θqεt-q+εt (48)
Wherein epsilontFor a white noise sequence, this formula embodies the time sequence xtThe autocorrelation function and the partial autocorrelation function of the model are respectively nonzero after the order of p and q, namely, the model has trailing property.
3. Modeling a photovoltaic generator set;
the power generated by a photovoltaic power generation module depends on three parameters, namely solar irradiance, site ambient temperature, and finally the characteristics of the module itself. Solar radiation was modeled as follows:
Figure GDA0003012943650000171
(49) in the formula: s is solar irradiance (kw/m)2) And α and β are parameters of the beta probability distribution function.
Ppv(s)＝N×FF×V(s)×I(s) (50)
Figure GDA0003012943650000172
V(s)＝Voc-Kv×Tc (52)
I(s)＝Sa×[Isc+Ki(Tc-25)] (53)
Figure GDA0003012943650000173
(50) In formulae (1) to (54): t iscIs the cell temperature, T, in degrees CelsiusAIs based on DEG CAmbient temperature of the bit, KvAnd KiThe temperature coefficients of voltage and current are V/DEG C, A/DEG C and NOTIs the nominal operating temperature of the photovoltaic cell, in degrees C, FF is the charge factor, IscIs a short-circuit current in the current characteristic, VocIs an open circuit voltage in the voltage characteristic, IMPPAnd VMPPIs the current and voltage at the maximum power point in the current and voltage characteristics; saIs the average solar irradiance.
Photovoltaic power generation P in the method of the inventionpvAnd (t, k) is obtained from historical measured data and is expressed as the generated power of the photovoltaic cell generator set under the condition of fully utilizing all solar irradiance in the kth hour.
4. Providing relevant constraint conditions of combined output of the wind-light-water combined system, and modeling the wind-light-water combined system;
4-1. wind, light and water combined system structure
According to the method disclosed by the invention, an optimized dispatching model comprising a wind-light-water combined unit and a conventional thermal power unit which are coordinated by a wind turbine generator set, a photovoltaic generator set and a pumped storage unit is established, wind power is utilized as much as possible on the basis of the principle of dispatching wind-light power generation preferentially, and the system structure is shown in fig. 1.
4-2. coordination mode of wind power and storage combined system
Considering that wind power generation and solar photovoltaic power generation have the characteristics of inaccurate prediction, randomness and instability, the wind power generation, the photovoltaic power generation and the pumped storage form a wind-light-water combined power generation system, and as can be seen from fig. 2, the specific coordination mode of the wind-light-water combined system is as follows: because the coordination period of the wind storage combined system scheduled in the day ahead can be effectively shortened after the pumped storage group is added, the method takes 24 hours a day as a cycle, and is divided into four periods per hour for 96 coordination periods. When the sum of the output of the wind generating set and the output of the photovoltaic generating set in a certain hour is larger than or equal to the average value of the output in the period, the hour is defined as a water pumping state, and the water pumping power is controlled in the stage to enable the time intervals in the hour to be in a radial systemThe wind and light power generation provided by the system is the same, and the redundant wind and light power generation is used for pumped storage and pumped output. When the sum of the output of the wind generating set and the output of the photovoltaic cell generating set in a certain hour is smaller than the average value of the predicted output of wind power, the hour is defined as a power generation state, the wind and light power generation at the stage is less, at the moment, water is discharged from a reservoir for power generation, and the power generation power is controlled to enable the electric quantity generated by the water pumping set in each period of the hour to be coordinated with the wind and light power generation quantity and then provide the same electric energy to the system. Therefore, the combined output value of the combined system in each pumping state and power generation state is obtained, and P for the combined power generation capacity of the wind-light energy storage combined power generation system is setws(t, k) represents a joint force output value at the kth hour (24 hours per cycle, four hours per hour).
4-3. wind, light and water combined system joint output constraint and other related constraints
1) The wind, light and water combined system ensures that the coordinated total output of each time period in an hour is kept stable, and the output active power of the wind power plant and the photovoltaic power generation is relatively smooth:
Figure GDA0003012943650000181
Pws(t,k)＝Pws(t,k+1) (56)
wherein:
Pws(t,k)＝Pt XX(t,k)+Pt YY(t,k) (57)
Pp(t,k)＝(Pw(t,k)+Ppv(t,k))Y(t,k)-Pt Y (58)
Pg(t,k)＝Pt X-(Pw(t,k)+Ppv(t,k))X(t,k) (59)
X(t,k)+Y(t,k)＝1 (60)
Figure GDA0003012943650000191
in the formula: pws(t, k) is the wind-solar-storage combined power generation power in the kth time period of the tth hour; pt Y,Pt XThe wind-solar energy storage combined system is used for generating combined power in a water pumping state and a power generation state respectively; pw(t, k) is the generating power of the wind turbine generator in the kth time period of the tth hour; ppv(t, k) is the generating power of the photovoltaic unit in the kth time period of the tth hour; pp(t, k) and Pg(t, k) respectively representing the pumping power and the generating power of the pumped storage unit in the kth time period of the tth hour; pwp.tThe sum of the generated power of the wind turbine generator set and the photovoltaic generator set in the tth hour; t is a period, namely 24 hours; k is the number of time segments per hour, i.e. 4 time segments.
3) And (3) output constraint of the wind-light-storage combined system:
Pmin(t,k)≤Pws(t,k)≤Pmax(t,k) (62)
(62) in the formula: pmin(t,k)、PmaxAnd (t, k) is the minimum value and the maximum value of power transmitted to the power grid by the wind-solar-storage combined system in the kth period in the tth hour.
4) The wind-solar storage combined system is restricted in pumping power and generating power:
Figure GDA0003012943650000192
Figure GDA0003012943650000193
0≤PP(t,k)≤Ppmax(t,k) (65)
in the formula: etagGenerating efficiency for the water pump; ppmax(t, k) and Pgmax(t, k) are respectively the upper limit of the pumping power and the upper limit of the generating power; e (t, k) is the storage capacity of the pumped storage power station; the interval between two adjacent periods is Δ T.
5) Wind power-actual generated power constraint of photovoltaic generator set:
Pw.min(t,k)≤Pw(t,k)≤Pw.max(t,k) (66)
(66) in the formula: pw.min(t, k) and Pw.maxAnd (t, k) is the minimum value and the maximum value of the installed capacity of the wind turbine generator in the kth time period of the tth hour.
PPv.min(t,k)≤Ppv(t,k)≤PPv.max(t,k) (67)
(67) In the formula: ppv.min(t, k) and Ppv.maxAnd (t, k) is the minimum value and the maximum value of the generated power of the photovoltaic generator set in the kth period of the tth hour.
5. Modeling the economic optimization operation of the power distribution network by using the constraint conditions in the step 4;
generally, wind power and photovoltaic power should not be wasted according to national energy policy, i.e. incorporating the requirements of all clean energy. And as the wind and light renewable energy sources which are preferentially scheduled are merged into the power grid, the output intermittence and the fluctuation of the thermal power generating unit are poorer and poorer, so that the cost of the whole power distribution network system is increased. In order to reduce the consumption of non-renewable energy sources and reduce the intermittence and fluctuation of the output of the thermal power generating unit, the renewable energy sources are maximally consumed, and the system cost of the power distribution network is reduced as much as possible. Output power P of single wind turbineIFThe approximate piecewise function of the relation v with the wind speed shows that the method provided by the invention maximizes the total output of the wind-light storage combined system in the period on the basis of meeting the requirement that the combined output of the wind-light storage combined system varies with the variation of the load requirement, so that the minimum output and the more stable output of the thermal power generating unit can be realized. According to experience, the investment cost of the thermal power generating unit occupies a large proportion, so that the strategy can reduce the cost of the whole power distribution system.
1) Objective function
The optimization operation model objective function of the power distribution network consists of two parts, namely construction cost and energy consumption cost, and the concrete model is as follows:
Min:Ctotal＝CM+CQ+CC+CR (68)
(68) in the formula: ctotalFor the total cost of the distribution network system, CMFor the cost of coal for thermal power generating units, CQFor exhaust gas emission costs, CCFor each in the power supply network systemThe operation and maintenance costs of the equipment. CRThe construction cost of the whole life cycle of the power distribution network system is balanced to the cost of each year.
2) Optimizing constraints
In order to ensure the stable operation of the power distribution network, the system needs to meet the demand and supply balance.
The system needs to satisfy the following energy balance equation:
PG(t,k)+Pws(t,k)＝yL(t,k) (69)
(69) in the formula: y isLThe user load is a continuous and derivable function in one period for the electricity demand of the user load at the kth moment in the tth hour; pws(t, k) is the power generation power of the wind-light-water combined system at the kth moment in the tth hour, and is a continuous derivative function in a period; pGAnd (t, k) is the power generated by the thermal power generating unit at the kth moment in the tth hour.
6. Randomly extracting sample points by using Monte-Carlo simulation to obtain corresponding system variables and construct an initial sample library;
and 5, knowing that the power distribution network consists of a wind-solar-energy storage combined system and a thermal power generating unit. On the premise of preferentially scheduling renewable energy sources, and an optimized output scheme of the wind-light-storage combined unit can be obtained in the model established in step 5, so that the optimal output P of the wind-light-storage combined unit at the kth hour in one period is selected in the methodwsAnd (t, k) is a random variable, and the value range of the variable satisfies the formula (56).
The method of the invention adopts the random simulation technology to randomly extract sampling points for a coordination time interval every 15 minutes, so that the extracted sample points can be uniformly distributed in the whole sampling space with equal probability. The obtained initial sample library also provides important foundation and guarantee for the construction and the correction of a subsequent gradient enhanced Kriging model.
Selecting the total generated power P in the periodws.nAs system variables:
Figure GDA0003012943650000211
the plane formed by the output of the wind-solar-energy-storage combined unit at each moment in a period is uniformly divided into 96 subintervals, n sample points are randomly extracted in each subinterval by utilizing a random simulation technology, 96 multiplied by n random variables are collected to form an initial sample library Z, and the expression is as follows:
Figure GDA0003012943650000212
carrying out calculation analysis on 96 multiplied by n initial samples selected randomly to obtain the total generated power P in n periodsws.nAs the system variable, there is,
S＝[X1,…Xn] (72)
by using the unit models, the energy balance equation and the unit output constraint conditions mentioned above and combining a BCC (bacterial population chemotaxis algorithm) optimization algorithm, a program for solving the MINLP is written under a Matlab platform to solve the power distribution network optimization operation model, and the most economic total cost of the power distribution network under n corresponding design schemes can be obtained as corresponding function values.
YS＝[Y1,…Yn]＝[Y1(X1),…Yn(Xn)] (73)
By the analysis, a two-dimensional sampling initial particle library is formed by taking the total generating power in the period of the wind-light-storage combined unit and the optimized total system cost as coordinate values respectively and recorded as (N)1,N2,…Nn) And the coordinates of the nth particle are recorded as Nn(Xn,Yn) Wherein X isnRepresenting the wind-solar total generated power P of the nth sample particlews.n，YnRepresenting the most economical total cost of the distribution network under the configuration scheme corresponding to the nth sample particle.
7. Constructing a gradient enhanced dynamic Kriging model, adding gradient information, and screening to obtain a new sample library;
in order to meet the requirement that the value of the system variable proposed in the step 5 is maximum on the basis of changing along with the change of the user load demand, the method improves the accuracy of the kriging model by introducing gradient information, and evolves to be a new proxy model method, namely a gradient enhanced kriging model. The model improves the accuracy of the proxy model by introducing first-order partial derivative information.
The specific analysis is as follows:
in order to screen an initial sample library, n × m design variables and corresponding n × m partial derivative values under a wind-light-storage combined output curve are selected near random variables; the information of the sample library after screening and the partial derivative thereof is as follows:
Figure GDA0003012943650000221
Figure GDA0003012943650000222
Figure GDA0003012943650000231
the gradient enhanced Kriging model constructed by the screened sample library has excessive sample points, and needs to add gradient information to screen characteristic sample points to optimize the model, so that the characteristic sample points meeting the gradient information need to be screened continuously according to constraint conditions to modify the Kriging model to improve the accuracy of the model. And obtaining an optimal sample library after screening, and optimizing a gradient enhanced dynamic Kriging model by using the optimal sample library to finally obtain a relatively accurate global optimal solution.
The selection of the characteristic sample points in the method is mainly based on the following gradient information:
the satisfaction at each sampling location varies with the variation in load demand according to the random variables above.
The change formula of the load in each sampling interval is as follows:
Figure GDA0003012943650000232
screening partial derivative information of each column of the formula (35) by using a formula (36), removing sample information which does not meet the conditions, and constructing a new optimal sample library SN。
SN＝(X1,X2…XN) (78)
8. And (3) obtaining a wind-solar-storage combined output scheme under N corresponding design schemes by using the unit models, the energy balance equations and the unit output constraint conditions mentioned in the step (4) and the step (5) and by using the optimal sample library in the step (7), writing a program for solving MINLP (bacterial population chemotaxis algorithm) under a Matlab platform by combining a BCC (body control programming) optimization algorithm, and obtaining an optimal economic energy utilization scheme and system operating cost of the power distribution network by optimizing simulation calculation. Storing and establishing optimal particle library
Figure GDA0003012943650000233
And comparing with the initial particle library constructed in the step 6 to verify the effectiveness of the strategy provided by the invention.

Claims (1)

1. A power distribution network economic optimization scheduling method based on an improved dynamic kriging model is characterized by comprising the following steps: the method comprises the following steps:
step 1, modeling the state of a pumped storage unit; the specific process is as follows:
determining the running state of the pumped storage unit according to the predicted value of the day-ahead wind power output, introducing a 0-1 integer variable to consider two stages of pumping and generating of the unit, and establishing a state model of the pumped storage unit as follows:
X(t,k)+Y(t,k)＝1 (1)
(1) in the formula: when X (t, k) is 1, the unit is a power generation stage in the kth period of the t hour, the coordination period is 24 hours, four periods are divided every hour, and k is 1,2,3, 4; when Y (t, k) is 1, the unit is a water pumping stage;
step 2, modeling output and user load of the wind turbine generator, wherein the specific process is as follows:
2.1 modeling the output of the wind turbine
1) Output characteristic of single wind turbine
According to the power characteristic curve of the wind turbine generator, the output power P of a single wind turbine generatorIFThe relationship v to wind speed can be approximated by a piecewise function of:
Figure FDA0003012943640000011
(2) in the formula: v. ofrAnd PWrRespectively representing the rated wind speed and the rated power of the fan in the kth time period of the tth hour; v. ofinTo cut into the wind speed; v. ofoutCutting out the wind speed;
2) describing characteristics of total output curve of wind turbine generator
Total generated power P of wind turbineWCan be expressed as:
Pw(t,k)＝efNW(t,k)PIF(t,k) (3)
0≤PIF(t,k)≤PWR (4)
(3) formulae (I) and (4): pIF(t,k),Pw(t, k) respectively representing the generating power of a single fan and the total wind turbine generator in the kth period of the tth hour, wherein the unit is MW; e and f are respectively the transmission efficiency and the power generation efficiency of the fan, and are expressed in percent; n is a radical ofW(t, k) is the number of fans which normally operate in the wind power plant in the kth time period in the tth hour; pWR(t, k) is rated power of each unit, and the unit is MW;
2.2 user load modeling
Modeling a user load by adopting a simple exponential smoothing model and an ARIMA model in a time sequence prediction module, trying to establish a mathematical model according to historical data of the load, and then establishing a mathematical expression of load prediction on the basis of the mathematical model to predict future loads;
simple exponential smoothing model
The simple model is one of exponential smoothing models, is a nonlinear estimation method, and has the basic principle of minimizing the mean square error between a predicted value and an observed value; the method is applicable to time series data that is free of trending and seasonal components;
the basic prediction formula is:
St＝Ayt+(1-A)St-1 (5)
the general predictive formula is:
St＝Ayt+(1-A)yt-1+…+(1-A)t-2y2+(1-A)t-1y1 (6)
(5) formulae (I) and (6): y istIs the observed data at time t; stThe smoothed data; a is a real number between 0 and 1;
the ARIMA model is a most widely used time series prediction model, and can analyze time series data containing seasonal components, and comprises 3 main parameters: the method comprises the following steps of (1) carrying out autoregressive order p, differential order d and moving average order q, wherein the general model is recorded as ARIMA (p, d and q);
difference
When the model is used, firstly, the time series is stabilized through difference, and the difference is divided into general difference and seasonal difference;
the general difference formula is:
Figure FDA0003012943640000031
wherein y istIs the original time sequence, B is the delay operator,
Figure FDA0003012943640000032
in order to be a first-order difference,
Figure FDA0003012943640000033
is a d-order difference;
the seasonal difference formula is:
Figure FDA0003012943640000034
wherein: y istIs a sequence with a period of T,
Figure FDA0003012943640000035
representing a seasonal difference operator;
② autoregressive moving average model ARMA (p, q)
Xt＝φ1Xt-1+φ2Xt-2+…φpXt-p+θ1εt-1+θ2εt-2+…+θqεt-q+εt (9)
Wherein epsilontFor white noise sequence, equation (9) represents time sequence XtThe correlation between t and p moments before t and q white noise sequences, the autocorrelation function and the partial autocorrelation function of the model are respectively nonzero after the order of p and q, namely, the model has trailing property;
step 3, modeling the photovoltaic generator set, wherein the concrete solving process is as follows:
the generated power of a photovoltaic power generation module depends on three parameters: solar irradiance, field ambient temperature, and the characteristics of the module itself; solar radiation is modeled in the manner described below:
Figure FDA0003012943640000036
(10) in the formula: s is solar irradiance (kw/m)2) α and β are parameters of the beta probability distribution function;
Ppv(s)＝N×FF×V(s)×I(s) (11)
Figure FDA0003012943640000037
V(s)＝Voc-Kv×Tc (13)
I(s)＝Sa×[Isc+Ki(Tc-25)] (14)
Figure FDA0003012943640000041
(11) [ formula (15) ]: t iscCell temperature, T, in degrees CAIs the ambient temperature in degrees C, KvAnd KiThe temperature coefficients of voltage and current are V/C, A/C, NOTIs the nominal operating temperature of the photovoltaic cell, with the unit C DEG, FF being the charge factor, IscIs a short-circuit current in the current characteristic, VocIs an open circuit voltage in the voltage characteristic, IMPPAnd VMPPIs the current and voltage at the maximum power point in the current and voltage characteristics; saIs the average solar irradiance; ppv(s) represents photovoltaic power generation power per unit solar irradiance;
photovoltaic power generation Ppv(t, k) is obtained by actually measured data and is expressed as the power generation power of the photovoltaic cell generator set under the condition of fully utilizing all solar irradiance in the kth time period of the tth hour;
step 4, providing relevant constraint conditions of combined output constraint of the wind-light-water combined system, and modeling the wind-light-water combined system; the specific implementation process is as follows:
4.1 wind-solar-water combined system modeling
Considering that wind power generation and solar photovoltaic power generation have the characteristics of inaccurate prediction, randomness and instability, the wind power generation, the photovoltaic power generation and pumped storage form a wind-light-water combined power generation system, and the wind-light-water combined system has the following specific coordination mode: because the coordination time interval of the wind-light storage combined system in day-ahead scheduling can be effectively shortened after the pumped storage unit is added, the 24-hour day period is taken as a period, four time intervals are divided every hour, and 96 coordination time intervals are provided; when the sum of the output of the wind generating set and the output of the photovoltaic generating set in a certain hour is larger than or equal to the average value of the output in the period, the hour is defined as a water pumping state, the wind and light generating capacity provided to the system in each time period in the hour is the same by controlling the water pumping power in the stage, and redundant wind and light power generation is used for pumping the water storage water pumping output; when a certain size is smallWhen the sum of the output of the wind generating set and the output of the photovoltaic cell generating set in the hour is smaller than the average value of the predicted output of the wind power, the hour is defined as a power generation state, the wind-solar power generation at the stage is less, at the moment, water is discharged from a reservoir for power generation, and the power generation power is controlled to ensure that the electric quantity generated by the water pumping set at each time interval in the hour is the same as the electric energy provided to the system after the coordination of the electric quantity generated by the wind-solar power generation; therefore, the combined output value of the combined system in each pumping state and power generation state is obtained, and P for the combined power generation capacity of the wind-light energy storage combined power generation system is setws(t, k) represents a joint force output value at the kth time at the tth hour in one cycle;
4.2 associated constraint of combined output constraint of wind, light and water combined system
1) The wind power storage combined system ensures that the coordinated total output of each time period in an hour is kept stable, and the output active power of the wind power plant is relatively smooth:
Figure FDA0003012943640000052
Pws(t,k)＝Pws(t,k+1) (17)
wherein:
Pws(t,k)＝Pt XX(t,k)+Pt YY(t,k) (18)
Pp(t,k)＝(Pw(t,k)+Ppv(t,k))Y(t,k)-Pt Y (19)
Pg(t,k)＝Pt X-(Pw(t,k)+Ppv(t,k))X(t,k) (20)
X(t,k)+Y(t,k)＝1 (21)
Figure FDA0003012943640000051
in the formula: pws(t, k) is the kth time period wind-solar-storage combined power generation power；Pt Y,Pt XThe wind-solar energy storage combined system is used for generating combined power in a water pumping state and a power generation state respectively; pw(t, k) is the generating power of the wind turbine generator in the kth time period of the tth hour; ppv(t, k) is the power generated by the photovoltaic generator set in the kth time period of the tth hour; pp(t, k) and Pg(t, k) respectively representing the pumping power and the generating power of the pumped storage unit in the kth time period of the tth hour; pwp.tThe sum of the generated power of the wind turbine generator set and the photovoltaic generator set in the tth hour; t is a period, namely 24 hours; k is the number of time segments in each hour, namely 4 time segments;
2) wind-solar-storage combined output constraint:
Pmin(t,k)≤Pws(t,k)≤Pmax(t,k) (23)
(23) in the formula: pmin(t,k)、Pmax(t, k) is the minimum value and the maximum value of power transmitted to the power grid by the wind-solar-storage combined system in the kth period in the tth hour;
3) the wind-solar storage combined system is restricted in pumping power and generating power:
Figure FDA0003012943640000061
Figure FDA0003012943640000062
0≤PP(t,k)≤Ppmax(t,k) (26)
in the formula: etagGenerating efficiency for the water pump; ppmax(t, k) and Pgmax(t, k) are respectively the upper limit of the pumping power and the upper limit of the generating power; e (t, k) is the storage capacity of the pumped storage power station; the interval between two adjacent time intervals is delta T;
4) wind power-actual generated power constraint of photovoltaic generator set:
Pw.min(t,k)≤Pw(t,k)≤Pw.max(t,k) (27)
(27) in the formula：Pw.min(t, k) and Pw.max(t, k) is the minimum value and the maximum value of the installed capacity of the wind turbine generator in the kth time period of the tth hour;
PPv.min(t,k)≤Ppv(t,k)≤PPv.max(t,k) (28)
(28) in the formula: ppv.min(t, k) and Ppv.max(t, k) is the minimum value and the maximum value of the generated power of the photovoltaic generator set in the kth time period in the tth hour;
and 5, modeling the economic optimization operation of the power distribution network by using the constraint conditions in the step 4, wherein the specific implementation process is as follows:
in order to reduce the consumption of non-renewable energy sources and the output intermittence and fluctuation of a thermal power generating unit, the renewable energy sources are maximally consumed, and the system cost of a power distribution network is reduced as much as possible; output power P of single wind turbineIFThe approximate piecewise function of the relation v with the wind speed can know that the total output of the wind-light storage combined system in the period is the maximum on the basis of meeting the requirement that the combined output of the wind-light storage combined system changes along with the change of the load requirement, so that the minimum output and the more stable output of the thermal power generating unit can be realized; according to experience, the investment cost of the thermal power generating unit occupies a large proportion, so that the cost of the whole power distribution system can be reduced;
1) objective function
The optimization operation model objective function of the power distribution network consists of two parts, namely construction cost and energy consumption cost, and the concrete model is as follows:
Min:Ctotal＝CM+CQ+CC+CR (29)
(29) in the formula: ctotalFor the total cost of the distribution network system, CMFor the cost of coal for thermal power generating units, CQFor exhaust gas emission costs, CCFor operating and maintaining the equipment in the power supply network system, CRThe construction cost of the whole life cycle of the power distribution network system is balanced to the cost of each year;
2) optimizing constraints
In order to ensure the stable operation of the power distribution network, the system needs to meet the demand and supply balance;
the system needs to satisfy the following energy balance equation:
PG(t,k)+Pws(t,k)＝yL(t,k) (30)
(30) in the formula: y isL(t, k) is the required electric quantity of the user load at the kth moment in the tth hour, and the user load is a continuous and derivable function in a period; pws(t, k) is the power generation power of the wind-light-water combined system at the kth moment in the tth hour, and is a continuous derivative function in a period; pG(t, k) is the power generation power of the thermal power generating unit at the kth moment in the tth hour;
step 6, randomly extracting sample points by using Monte-Carlo simulation to obtain corresponding system variables, and constructing an initial sample library; the specific implementation process is as follows:
selecting optimal output P of the wind-light-storage combined unit at the kth moment in the tth hour in one periodws(t, k) are random variables, and the value range of the variables meets the wind-solar-storage combined output constraint Pmin(t,k)≤Pws(t,k)≤Pmax(t,k)；
Sampling points are randomly extracted for a coordination time interval every 15 minutes by adopting a random simulation technology, so that the extracted sample points can be uniformly distributed in the whole sampling space at equal probability; the obtained initial sample library also provides important basis and guarantee for the construction and the correction of a subsequent gradient enhanced Kriging model;
selecting the total generated power P in the periodws.nAs system variables:
Figure FDA0003012943640000081
the plane formed by the output of the wind-solar-energy-storage combined unit at each moment in a period is uniformly divided into 96 subintervals, n sample points are randomly extracted in each subinterval by utilizing a random simulation technology, 96 multiplied by n random variables are collected to form an initial sample library Z, and the expression is as follows:
Figure FDA0003012943640000082
carrying out calculation analysis on 96 multiplied by n initial samples selected randomly to obtain the total generated power P in n periodsws.nAs a function of the system variables,
S＝[X1,…Xn] (33)
compiling a MINLP solving program under a Matlab platform by using the unit models, the energy balance equations and the unit output constraint conditions and combining a BCC (bacterial colony chemotaxis algorithm) optimization algorithm to solve the economic optimization operation model of the power distribution network, so that the most economic total cost of the power distribution network under n corresponding design schemes can be obtained as corresponding function values;
YS＝[Y1,…Yn]＝[Y1(X1),…Yn(Xn)] (34)
respectively taking the total generating power in the period of the wind-light-storage combined unit and the total cost of the optimized system as coordinate values to form a two-dimensional initial particle library Q which is recorded as (N)1,N2,…Nn) And the coordinates of the nth particle are recorded as Nn(Xn,Yn) Wherein X isnRepresenting the wind-solar total generated power P of the nth sample particlews.n，YnRepresents the most economic total cost of the distribution network under the configuration scheme corresponding to the nth sample particle, NnThe system composition scheme representing the power distribution network can be recorded as scheme Nn；
Step 7, constructing a gradient enhanced dynamic Kriging model, and adding gradient information to screen an initial sample library; the specific implementation process is as follows:
in order to meet the requirement that the value of a system variable is maximum on the basis of changing along with the change of the load requirement of a user, the accuracy of the kriging model is improved by introducing gradient information, and the kriging model is evolved into a new agent model method, namely a gradient enhanced kriging model; the accuracy of the proxy model is improved by introducing first-order partial derivative information; the specific analysis is as follows:
in order to screen an initial sample library, selecting n × m design variables and corresponding n × m partial derivative values under a wind-light-storage combined output curve near the random variables; the information of the sample library after screening and the partial derivative thereof is as follows:
Figure FDA0003012943640000091
Figure FDA0003012943640000092
Figure FDA0003012943640000093
in the gradient enhanced Kriging model constructed by the screened sample library, because the number of sample points is too large, gradient information is required to be added to screen characteristic sample points to optimize the model, so that the characteristic sample points meeting the gradient information need to be screened according to constraint conditions to modify the Kriging model so as to improve the accuracy of the model; obtaining the optimal sample library S after screeningNReuse the sample library SNOptimizing a gradient enhanced dynamic Kriging model to finally obtain a relatively accurate global optimal solution;
the selection of the characteristic sample points is mainly based on the following gradient information:
the random variable is changed along with the change of the load requirement at each sampling position; the partial derivative change formula of the load in each sampling interval is as follows:
Figure FDA0003012943640000094
screening partial derivative information of each column of the formula (37) by using a formula (38), removing sample information which does not meet the conditions, and constructing an optimal sample library SN：
SN＝(X1,X2…XN)； (39)
And 8, solving the economic optimization model of the power distribution network by using a BCC optimization algorithm on the basis of the new sample library, wherein the specific implementation process is as follows:
by utilizing the unit models, the energy balance equations and the unit output constraint conditions mentioned in the step 4 and the new sample library in the step 7, the wind-solar-storage combined output scheme under N corresponding design schemes can be obtained, a program for solving MINLP is written under a Matlab platform by combining a BCC optimization algorithm, and the optimal economic energy utilization scheme and the system operation cost of the power distribution network can be obtained through optimization simulation calculation; storing and establishing optimal particle library
Figure FDA0003012943640000101
And comparing with the constructed initial particle library in the step 6 to verify the effectiveness of the strategy provided by the invention.
CN201811057761.5A 2018-09-11 2018-09-11 Power distribution network economic optimization scheduling method based on improved dynamic kriging model Active CN109103929B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811057761.5A CN109103929B (en) 2018-09-11 2018-09-11 Power distribution network economic optimization scheduling method based on improved dynamic kriging model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811057761.5A CN109103929B (en) 2018-09-11 2018-09-11 Power distribution network economic optimization scheduling method based on improved dynamic kriging model

Publications (2)

Publication Number Publication Date
CN109103929A CN109103929A (en) 2018-12-28
CN109103929B true CN109103929B (en) 2021-07-20

Family

ID=64865903

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811057761.5A Active CN109103929B (en) 2018-09-11 2018-09-11 Power distribution network economic optimization scheduling method based on improved dynamic kriging model

Country Status (1)

Country Link
CN (1) CN109103929B (en)

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109713734B (en) * 2019-02-15 2019-12-06 西华大学 Photovoltaic power adjusting method, device, equipment and medium
CN110110385B (en) * 2019-04-12 2022-05-03 电子科技大学 Application method of composite-based adaptive agent model in battery module optimization design
CN110135631B (en) * 2019-04-26 2022-02-22 燕山大学 Electric comprehensive energy system scheduling method based on information gap decision theory
CN110391677B (en) * 2019-08-26 2022-11-22 电子科技大学 Water-light storage hybrid system operation optimization method based on electric power market environment
CN111064229B (en) * 2019-12-18 2023-04-07 广东工业大学 Wind-light-gas-storage combined dynamic economic dispatching optimization method based on Q learning
CN112149960A (en) * 2020-08-27 2020-12-29 国网浙江海宁市供电有限公司 Photo-electricity-micro coordination control method based on prediction of photovoltaic power generation
CN112906251B (en) * 2021-04-16 2023-05-02 云南电网有限责任公司 Analysis method and system for reliability influence elements of power distribution network
CN115481791A (en) * 2022-09-05 2022-12-16 中国长江三峡集团有限公司 Water-wind power generation and power generation combined prediction method, device and equipment

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102063557A (en) * 2010-09-07 2011-05-18 合肥兆尹信息科技有限责任公司 Method and system for forecasting data
WO2011062794A1 (en) * 2009-11-18 2011-05-26 Conocophillips Company Attribute importance measure for parametric multivariate modeling
CN103136428A (en) * 2013-03-12 2013-06-05 上海交通大学 Vehicle body structure steady design method based two uncertain saloon cars
CN106981888A (en) * 2017-05-10 2017-07-25 西安理工大学 The multiple target dynamic dispatching method of Thermal and Hydroelectric Power Systems is stored based on the complementary wind of multi-source
CN107622324A (en) * 2017-09-01 2018-01-23 燕山大学 A kind of robust environmental economy dispatching method for considering more microgrid energy interactions

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2011062794A1 (en) * 2009-11-18 2011-05-26 Conocophillips Company Attribute importance measure for parametric multivariate modeling
CN102063557A (en) * 2010-09-07 2011-05-18 合肥兆尹信息科技有限责任公司 Method and system for forecasting data
CN103136428A (en) * 2013-03-12 2013-06-05 上海交通大学 Vehicle body structure steady design method based two uncertain saloon cars
CN106981888A (en) * 2017-05-10 2017-07-25 西安理工大学 The multiple target dynamic dispatching method of Thermal and Hydroelectric Power Systems is stored based on the complementary wind of multi-source
CN107622324A (en) * 2017-09-01 2018-01-23 燕山大学 A kind of robust environmental economy dispatching method for considering more microgrid energy interactions

Non-Patent Citations (6)

* Cited by examiner, † Cited by third party
Title
Zhigang Lu, Liye ma.Power System Reactive Power Optimization Based on Direct Neural Dynamic Programming.《Proceedings of 2008 3rd International Conference on Intelligent System and Knowledge Engineering》.2008,全文. *
丁荣荣.配电网经济运行综合评价及灵敏度分析.《中国优秀硕士学位论文全文数据库》.2017,全文. *
沈洲.基于机会约束规划和随机模拟技术的含风电场电力系统多目标优化调度.《电力学报》.2013,第28卷(第1期),全文. *
王振浩.考虑风电消纳的风电–电储能–蓄热式电锅炉联合系统能量优化.《中国电机工程学报》.2017,第37卷全文. *
钟嘉庆.基于多场景和区间模糊数的输电网规划综合决策.《系统工程理论与实践》.2013,第33卷(第9期),全文. *
韩忠华.Ｋｒｉｇｉｎｇ模型及代理优化算法研究进展.《航空学报》.2016,第37卷(第11期),全文. *

Also Published As

Publication number Publication date
CN109103929A (en) 2018-12-28

Similar Documents

Publication Publication Date Title
CN109103929B (en) Power distribution network economic optimization scheduling method based on improved dynamic kriging model
CN105375479B (en) A kind of distributed energy energy management method based on Model Predictive Control
CN103151803B (en) Method for optimizing wind power system-contained unit and backup configuration
CN107134810B (en) Independent micro-energy-grid energy storage system optimal configuration solving method
CN110365013B (en) Capacity optimization method of photo-thermal-photovoltaic-wind power combined power generation system
CN112583017B (en) Hybrid micro-grid energy distribution method and system considering energy storage operation constraint
CN112736926A (en) Interval affine power flow dynamic optimization method for distributed new energy access power distribution network
CN112907098B (en) Multi-stage capacity configuration method and configuration system of park comprehensive energy system
CN113850474B (en) Thermoelectric hydrogen multi-energy flow comprehensive energy system and optimal scheduling method thereof
Li et al. Optimized operation of hybrid system integrated with MHP, PV and PHS considering generation/load similarity
CN110601260A (en) Light-storage system capacity optimization method for limiting power fluctuation on interconnection line
CN110867907B (en) Power system scheduling method based on multi-type power generation resource homogenization
CN116402210A (en) Multi-objective optimization method, system, equipment and medium for comprehensive energy system
CN116316694A (en) Energy storage power station frequency modulation optimal parameter selection method based on two-stage robust optimization
CN113435659B (en) Scene analysis-based two-stage optimized operation method and system for comprehensive energy system
CN114066204A (en) Integrated optimization planning and operation method and device of comprehensive energy system
CN112134307A (en) Electric energy storage/heat energy storage capacity joint optimization method and system of multi-energy complementary system
CN109546647B (en) Safety and stability evaluation method for power system containing wind, light and water storage
CN115907402B (en) Method and system for evaluating joint guaranteed output of cascade hydropower station
Li et al. Impact on traditional hydropower under a multi-energy complementary operation scheme: An illustrative case of a ‘wind–photovoltaic–cascaded hydropower plants’ system
CN116979611A (en) Hierarchical optimization scheduling method for source network load storage
CN114938040B (en) Comprehensive optimization regulation and control method and device for source-network-load-storage alternating current-direct current system
CN115271244A (en) Two-stage distribution robust optimization-based short-term peak regulation model of cascade hydropower station
CN114188942A (en) Power grid dispatching method comprising large-scale new energy base
CN114421536A (en) Multi-energy interactive regulation and control method based on energy storage

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant
TR01 Transfer of patent right
TR01 Transfer of patent right

Effective date of registration: 20230517

Address after: 2081, building a, 88 Jianghai West Road, Liangxi District, Wuxi City, Jiangsu Province, 214000

Patentee after: Wuxi Xiangyuan Information Technology Co.,Ltd.

Address before: 066004 No. 438 west section of Hebei Avenue, seaport District, Hebei, Qinhuangdao

Patentee before: Yanshan University