CN113536604B

CN113536604B - Structural health monitoring sensor layout optimization method for constraint non-gradient general calculation

Info

Publication number: CN113536604B
Application number: CN202111022746.9A
Authority: CN
Inventors: 尹涛
Original assignee: Wuhan University WHU
Current assignee: Wuhan University WHU
Priority date: 2021-09-01
Filing date: 2021-09-01
Publication date: 2022-08-05
Anticipated expiration: 2041-09-01
Also published as: CN113536604A

Abstract

The invention belongs to the technical field of structure detection and health monitoring, and discloses a structural health monitoring sensor layout optimization method for constraint gradient-free general solution, which comprises the steps of determining an optimized layout criterion of a target structural sensor, and selecting a gradient-free optimization algorithm with boundary constraint; and constructing a target function for optimizing the layout of the sensors, expanding a vector translation interface for optimizing the target function, and obtaining an optimized combination of the sensor layout by using constraint non-gradient general calculation. The method can effectively solve the problem of optimization and layout combination of the sensors with large structures, and can generally use all constrained gradient-free optimization algorithms including artificial intelligence and classic gradient-free optimization algorithms considering boundary constraint for solving the combination optimization problem of sensor optimization configuration by only expanding an optimization vector coding and translating interface of the objective function without changing the codes of the optimization algorithms.

Description

Structural health monitoring sensor layout optimization method for constraint non-gradient general calculation

Technical Field

The invention belongs to the technical field of structure detection and health monitoring, and particularly relates to a method and a system for optimizing the layout of a structural health monitoring sensor for constraint non-gradient general calculation.

Background

At present, a large civil engineering structure is comprehensively influenced by various static and dynamic loads, environmental erosion and other factors in a long service period, the technical condition and the use function of the large civil engineering structure are always degraded continuously, the use safety of the large civil engineering structure is reduced, and even the whole safety of the structure is threatened.

The structural health monitoring system generally comprises a data acquisition system, a transmission system, an analysis system, a decision-making system and the like, wherein the sensor system is an important component of the structural health monitoring data acquisition system, the number of used sensors and the specific arrangement scheme of the sensors on a structure to be monitored are critical to the quality of acquired measured data, and the success or failure of the structural health monitoring system based on the analysis and evaluation of the static and dynamic measurement information is directly influenced. However, at present, in practical applications, the arrangement of the sensor system is still mostly based on experience, i.e. the arrangement scheme of the sensor system on the structure to be monitored is artificially determined by considering a series of practical constraints, but this is not significant for structural health monitoring targeting identification of the structural damage degradation characterization parameters, because this empirical approach cannot ensure the sensitivity of the measurement information picked up by the sensor system to the damage parameters to be monitored. Therefore, the system develops the research on the optimal layout of the sensors, can ensure that the selected sensor configuration scheme can obtain enough structural static-dynamic response information quantity, and provides high-quality data support for accurately analyzing and judging the service state of the structure.

The main defects of the prior sensor optimal configuration problem resolving technology are as follows:

the solution of the sensor layout optimization problem in the structural health monitoring generally needs to satisfy a constraint condition with a specific physical meaning, that is, the total number of sensors actually available for layout is required to be kept unchanged all the time in the algorithm optimization iteration process, and how to effectively satisfy the constraint condition is always the focus of attention and the solution difficulty of the sensor optimization configuration combination optimization problem.

At present, the most common method is to directly modify the source code of the optimization algorithm to satisfy the mandatory constraint condition (such as papers DOI:10.1177/1475921719877579, DOI: 10.3969/j.issn.1000-4750.2000.01.004; patents CN105976018A, CN108537320A), wherein the genetic algorithm is a type of artificial intelligence optimization method which is currently researched most widely in the field of structural health monitoring sensor optimization layout and related scientific fields related to combinatorial optimization, the problem to be solved is not required to have a definite mathematical equation and a derivative expression, the global optimization capability is strong, and the search process does not depend on gradient information, so that the method is very suitable for solving the type of combinatorial optimization problem. The practical application of the method usually depends on the directional modification or redesign of various internal genetic operator codes to realize the solution of the constrained combined optimization process (such as papers DOI: 10.3321/j.issn:0372-2112.2007.10.034, DOI: 10.1260/0266-3511.29.3.121; patent CN112801387A), and the mandatory constraint condition of the sensor optimization configuration is met, so that the total number of the sensors is always kept unchanged in the algorithm evolution process. Nevertheless, in the specific algorithm implementation process for solving the combinatorial optimization problem, it is often necessary to spend considerable workload on determining the coding scheme and designing genetic operators (crossover, mutation, etc.) (see papers DOI: 10.3969/j. issn.1000-3835.2008.03.021, DOI: 10.1177/1475921719877579). It should be noted, however, that the above directed design and modification process of the optimization algorithm can only be generally specific to the specific algorithm under study, and the corresponding source code modification strategy needs to be specifically studied for different algorithms, but the process usually needs to deeply interpret the source codes of various optimization algorithms, and simultaneously understand and grasp the theoretical backgrounds of various algorithms and their corresponding program implementations in detail, and the process is heavy, cumbersome and error-prone. On the other hand, in the face of the effective solving requirement of the high-dimensional combinatorial optimization problem of the optimized layout of the large-scale structural sensor, the high efficiency and the reliability of the adopted optimization algorithm must be ensured, however, most of the existing optimization algorithms with both high efficiency and reliability are commercialized encapsulation programs or other mature closed-source programs, and the source code of the optimization algorithms cannot be obtained, so that the strategy of trying to satisfy the mandatory constraint condition of the sensor layout combinatorial optimization by directionally designing and modifying the algorithm source code is not feasible, and the strong advantages of the mature commercial closed-source codes on the calculation of the large-scale high-dimensional combinatorial optimization problem of the large-scale engineering structure cannot be effectively utilized and fully exerted.

Through the above analysis, the problems and defects of the prior art are as follows: the existing structural health monitoring sensor layout optimization technology relying on the directional modification or redesign optimization algorithm has the disadvantages of poor adaptability and universality, large workload, complexity and error easiness; meanwhile, the directional code modification strategy is not beneficial to practical large-scale engineering structure health monitoring and super-large-scale combined decision problem real scene application due to the fact that efficient and reliable program codes cannot be obtained.

Disclosure of Invention

Aiming at the problems in the prior art, the invention provides a structural health monitoring sensor layout optimization method and system for constraint non-gradient general calculation.

The invention is realized in such a way that a structural health monitoring sensor layout optimization method for constraint non-gradient general solution comprises the following steps:

determining an optimal layout criterion of a target structure sensor, and selecting a gradient-free optimization algorithm with boundary constraint;

and secondly, constructing a target function for optimizing the layout of the sensors, expanding a vector translation interface for optimizing the target function, and obtaining an optimized combination of the sensor layout by using constraint non-gradient general calculation.

Further, in the first step, the determining the optimal layout criterion of the target structure sensor includes:

analyzing a design drawing of a structure to be monitored and other related technical data, comprehensively considering the characteristics of a space structure system of the structure to be monitored, establishing a finite element numerical model of a target structure, and drawing out all possible sensor layout positions based on the model; and determining key monitoring positions of the finite element models by combining with the actual health monitoring requirements of the target structure, and constructing a layout optimization criterion of a sensor system by taking the measurement information requirements of quantitative inversion identification of the damage and degradation characterization parameters of the key positions as main targets in a probability theory and information theory framework, so that the M finite sensors arranged according to the criterion can obtain relatively most abundant static and dynamic force measurement data information, and the quantitative inversion information requirements of the damage and degradation parameters are met to the maximum extent.

Further, the total number of all possible sensor layout positions is N.

Further, in the first step, the gradient-free optimization algorithm with boundary constraint includes: the algorithm adopts continuous real number vector coding, and the upper and lower boundaries of each element in the unknown vector are respectively taken as 0 and 1.

Further, in step two, the constructing an objective function of the sensor optimized layout includes:

and (3) representing a sensor configuration scheme by a continuous real number optimization vector in a gradient-free optimization algorithm with boundary constraint, evaluating a utility function value of each sensor layout scheme according to the layout quasi-measured quantity, and constructing a sensor layout target function.

Furthermore, each element in the real number type optimization vector corresponds to a specific sensor arrangement position on the structure to be monitored respectively; the optimization vector total dimension N represents the total number of all possible sensor arrangement positions proposed.

Further, in step two, the extended objective function optimized vector translation interface includes:

(1) performing descending arrangement on the N-dimensional continuous real number optimization vectors; extracting the generated N-dimensional sequencing number vector; intercepting the first M number elements of the N-dimensional number vector;

(2) independently generating an N-dimensional 0 element vector; performing transformation operation on the generated N-dimensional 0 element vector according to the intercepted M-dimensional number vector, namely setting elements corresponding to M number values in the N-dimensional 0 element vector as 1 and keeping the rest elements unchanged, and completing the translation from the N-dimensional continuous vector to the N-dimensional 01 element vector;

(3) taking the obtained N-dimensional 01-type element vector as an indication vector of the sensor layout position, and evaluating a utility function value of a layout scheme corresponding to the sensor by using the indication vector;

(4) further taking the utility function value as a target function value actually corresponding to an original N-dimensional continuous real number vector outside the interface, and completing the utility evaluation process of continuous optimization vector translation coding;

(5) and based on the target function after the translation interface is expanded, effectively resolving the problem of optimizing the sensor layout and combination by using the constrained non-gradient algorithm to obtain a sensor layout and combination optimizing scheme.

It is another object of the present invention to provide a structural health monitoring method or a structural monitoring system implementing a structural health monitoring sensor layout optimization method that constrains the gradient-free general solution.

By combining all the technical schemes, the invention has the advantages and positive effects that: the method can effectively solve the large-scale high-dimensional combinatorial optimization problem of the optimized layout of the large-scale structural sensors, can generally use all constrained gradient-free optimization algorithms including artificial intelligence algorithms considering boundary constraints and classical gradient-free optimization algorithms to solve the combinatorial optimization problem of the optimized configuration of the sensors by only expanding an optimization vector coding translation interface of a target function under the condition of not changing codes of the optimization algorithms, and can fully utilize and exert the strong advantages of mature and reliable codes of the closed-source commercial optimization algorithms in the solution of the large-scale high-dimensional combinatorial optimization problem of the large-scale structural sensors.

The invention does not need to modify the gradient-free algorithm: the method does not need any directional design or local modification on the source code of the optimization algorithm, can completely use mature and reliable commercial closed-source codes as the black box, can fully utilize and exert the strong operational efficiency of various advanced optimization algorithms, and greatly facilitates the application efficiency of the advanced and efficient optimization algorithms in the solution of the optimization problem of the optimized layout and combination of large-scale sensors with large structures.

The invention is widely applicable to non-gradient optimization algorithms: the method can be generally suitable for all continuous real variable type gradient-free optimization algorithms considering upper and lower boundary constraints on the premise of not changing an algorithm source code, comprises artificial intelligent algorithms such as a traditional genetic algorithm, a particle swarm algorithm, an ant colony algorithm and the like, a classical Nelder-Mead simplex algorithm with boundary constraints and the like, and solves the optimization problem of the optimization layout combination of the large-scale structure high-dimensional sensors by efficiently utilizing the mature and reliable gradient-free algorithms.

The invention only carries out standardized extension on the objective function: the invention changes the source code modification of the optimization algorithm into the method of only adding a standardized expansion interface to the target utility function of the sensor optimization arrangement, namely, the real number codes of continuous vectors corresponding to all possible configuration positions of the sensor one by one can be translated into 01 coding type indication vectors reflecting the real sensor arrangement scheme, and the quantity information of the sensors is ensured to be kept constant in the translation process, namely, the quantity of the arranged sensors is kept constant, thereby naturally meeting the mandatory constraint condition of the optimization problem of the sensor optimization arrangement combination.

The invention can greatly improve the efficiency through algorithm parallelism: the invention can fully utilize the parallel computing or distributed computing strategy of the original non-gradient optimization algorithm, such as the parallel genetic algorithm, the parallel particle swarm algorithm, the parallel ant colony algorithm and the like, and obtain remarkable computing performance improvement in the solution of the large-scale high-dimensional combinatorial optimization problem of the optimization configuration of the large-scale structural health monitoring sensor, and in the process of the combinatorial optimization parallel computing, the objective function after the interface expansion does not need to be changed and adjusted any more, thereby ensuring the efficient implementation of the algorithm parallel.

Drawings

FIG. 1 is a flow chart of a method and a system for optimizing the layout of structural health monitoring sensors by constrained gradient-free general solution according to an embodiment of the present invention.

FIG. 2 is a flowchart of a sensor layout optimization objective function translation interface extension and optimization process for constrained gradient-free general solution according to an embodiment of the present invention.

Fig. 3 is a schematic diagram of a model of a large-span suspension bridge according to an embodiment of the present invention.

Fig. 4(a) is a schematic diagram of an iteration result of optimally laying 1 sensor according to an embodiment of the present invention.

Fig. 4(b) is a schematic diagram of iteration results of optimally laying 2 sensors according to the embodiment of the present invention.

Fig. 4(c) is a schematic diagram of an iteration result of optimally laying 3 sensors according to an embodiment of the present invention.

Fig. 4(d) is a schematic diagram of an iteration result of optimally laying 4 sensors according to an embodiment of the present invention.

Fig. 4(e) is a schematic diagram of an iteration result of optimally laying 5 sensors according to an embodiment of the present invention.

Fig. 4(f) is a schematic diagram of an iteration result of optimally laying 6 sensors according to an embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention is further described in detail with reference to the following embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

Aiming at the problems in the prior art, the invention provides a structural health monitoring sensor layout optimization method and system for constraint gradient-free general calculation, and the invention is described in detail below with reference to the accompanying drawings.

As shown in fig. 1, the method for optimizing the layout of a structural health monitoring sensor by constrained gradient-free general solution according to the embodiment of the present invention includes:

s101, determining an optimal layout criterion of a target structure sensor, and selecting a gradient-free optimization algorithm with boundary constraint;

s102, constructing a target function for optimizing the layout of the sensors, expanding a vector translation interface for optimizing the target function, and obtaining an optimized combination of the sensor layout by constraint non-gradient general calculation.

The method for determining the optimal layout criterion of the target structure sensor comprises the following steps:

The total number of all possible sensor layout positions provided by the embodiment of the invention is N.

The gradient-free optimization algorithm with boundary constraint provided by the embodiment of the invention comprises the following steps: the algorithm adopts continuous real number vector coding, and the upper and lower boundaries of each element in the unknown vector are respectively taken as 0 and 1.

The objective function for constructing the optimized layout of the sensor provided by the embodiment of the invention comprises the following steps:

Each element in the real number type optimization vector provided by the embodiment of the invention respectively corresponds to a specific sensor arrangement position on a structure to be monitored; the optimization vector total dimension N represents the total number of all possible sensor arrangement positions proposed.

The vector translation interface for optimizing the extended objective function provided by the embodiment of the invention comprises:

(4) further taking the utility function value as a target function value actually corresponding to the original N-dimensional continuous real number vector outside the interface, and finishing the utility evaluation process of the continuous optimization vector translation coding;

(5) and based on the target function after the translation interface is expanded, effectively solving the problem of optimizing the sensor layout optimization combination by using the constrained gradient-free algorithm to obtain a sensor layout optimization combination scheme.

The technical solution of the present invention is further described with reference to the following specific embodiments.

Example 1:

(1) determining the optimal layout rule of the target structure sensor: on the basis of researching and analyzing related technical data such as a structural design drawing to be monitored and the like, comprehensively considering the characteristics of a space structure system, establishing a finite element numerical model of a target structure, and drawing up all possible sensor arrangement positions on the basis of the finite element numerical model, wherein the total number of the possible sensor arrangement positions is set to be N; and further combining the finite element model and the actual health monitoring requirements of the target structure, reasonably determining a criterion for sensor layout optimization, so that M finite sensors arranged according to the criterion can acquire relatively abundant static and dynamic force measurement data information.

(2) Selecting a gradient-free optimization algorithm with boundary constraint: comprehensively comparing various available gradient-free optimization algorithms with boundary constraints from the aspects of algorithm global optimization capability, convergence speed, convenience and quickness in application, algorithm parallelism and the like, and selecting a proper gradient-free optimization algorithm with boundary constraints from upper and lower sides; in the algorithm, an N-dimensional continuous real number optimization vector is considered, and the upper boundary and the lower boundary of each element in an unknown vector are respectively taken as 0 and 1, so that each optimization variable can be continuously changed between 0 and 1 in the algorithm optimization process, and continuous optimization conditions are met.

(3) Constructing an objective function of optimal layout of the sensor: representing a sensor configuration scheme through a continuous real number optimization vector in the non-gradient optimization algorithm in the step (2), wherein each element in the real number optimization vector corresponds to a specific sensor arrangement position on a structure to be monitored respectively, and the total dimension N of the optimization vector represents the total number of all possible sensor arrangement positions proposed in the step (1); and (3) evaluating the utility function value of each sensor layout scheme according to the layout accuracy measurement quantity in the step (1) so as to construct a sensor layout target function, wherein the optimized vector of the sensor layout target function is a continuous real number and can be utilized by the algorithm in the step (2) after passing through the expansion interface.

(4) The extended objective function optimizes the vector translation interface: and (4) on the basis of the optimal layout target function of the structural sensor to be monitored constructed in the step (3), expanding an interface for performing optimal vector translation on the target function by applying the method. The method can be particularly subdivided into the following main steps: carrying out descending order arrangement on the N-dimensional continuous optimization vectors; extracting N-dimensional sequencing number vectors generated in the first step; thirdly, intercepting the first M elements of the N-dimensional numbered vector in the second step; fourthly, independently generating an N-dimensional 0 element vector; fifthly, carrying out transformation operation on the N-dimensional 0 vector generated in the fourth step according to the M-dimensional number vector intercepted in the third step, namely setting elements corresponding to the M number values in the N-dimensional 0 vector as 1 and keeping the rest elements unchanged, and finishing the translation from the continuous vector to the 01 vector; sixthly, taking the N-dimensional 01 vector obtained in the fifth step as a sensor layout indication vector which can be directly adopted by the target function in the step (3), and effectively solving the problem of sensor optimization configuration combination optimization based on the constrained gradient-free algorithm in the step (2). It is worth pointing out that the translation extension interface constructed by the present invention can seamlessly connect the boundary-constrained gradient-free optimization algorithm in step (2) with the target function for sensor optimization configuration of the object to be monitored constructed in step (3) very effectively, and does not need any modification to the original optimization algorithm code.

Example 2:

in order to verify the feasibility and the calculation efficiency of solving the problem of optimal layout of the sensors based on the boundary constraint non-gradient optimization algorithm expanded by the invention, a large-span suspension bridge model shown in fig. 2 is considered, and a plurality of uniaxial acceleration sensors are optimally arranged at all possible nodes of the bridge floor, wherein the number of the possible layout positions is N-70, and the possible layout positions are identified by circles shown in the figure. In this case, a total of 6 different numbers of sensors are considered for the installation of the bridge deck, i.e. 1 to 6 sensors are respectively arranged in 70 possible positions in 6 cases. As a key force transmission component of a suspension bridge structure system, the self weight and the external load of a bridge deck stiffening girder are transmitted to a main cable by a sling to become an important tie for connecting the bridge deck stiffening girder and the main cable, and the health service state of the sling plays a decisive role in the safety of the whole bridge structure system. Specifically, every pair of slings on the same section of the upstream and downstream of the suspension bridge model are divided into one group, and 33 groups of slings are totally distributed on two side spans and a main span of the bridge. The method comprises the steps of representing the health state of the sling according to the elastic modulus of the sling, calibrating or correcting the elastic modulus of each group of slings according to 33 dimensionless proportionality coefficients, and carrying out inversion identification on the dimensionless proportionality parameters reflecting the health state of the sling by using effective modal parameter data obtained by a dynamic test. The modal parameter information quantity acquired from the specified sensor layout scheme is quantitatively represented through an information entropy theory and is used as a target function for optimizing layout of the sensors, namely, a group of sensor optimization layout schemes are searched on the basis of all 70 possible layout positions of the bridge floor, so that the sensor combination laid according to the schemes can measure relatively most abundant modal parameter information, and effective identification of the cable health state representation dimensionless coefficient based on the bridge floor modal parameters is facilitated to be carried out subsequently.

Without loss of generality, the embodiment takes a genetic algorithm, which is a typical representative in a gradient-free optimization algorithm, as an example for application research, and specifically selects a real number encoding genetic algorithm considering boundary constraints, and it should be pointed out that the invention can completely use the existing general and closed-source genetic algorithms as black boxes, does not need to change any genetic operator or any code inside the genetic algorithm, and only needs to simply extend a specific sensor optimization layout objective function by applying the invention to construct a seamless interface for performing optimization variable translation with the traditional black box genetic algorithm, so that the problem of optimization and combination of sensor optimization configuration can be effectively solved on the basis of the traditional genetic algorithm, and a global optimal configuration combination can be found with certain possibility under the condition of ensuring sufficient population scale. On the basis, the resolving efficiency of large-scale high-dimensional combination optimization problems such as large-scale bridge structure sensor configuration and the like can be obviously improved by further utilizing the parallel or distributed computing function of the genetic algorithm.

The computer adopted in the embodiment is configured as a Windows 1064-bit professional edition, i7-6820HQ CPU @2.70GHz, a 32G memory and a 512G hard disk, a computing software platform is MATLAB R2019a, and a parallel computing function is considered, wherein a Genetic Algorithm adopts a built-in standard Algorithm in a Genetic Algorithm Toolbox of the software platform, the population scale of the Genetic Algorithm is taken as a software default value, namely 10 times of the number of unknown variables (in the example, the number of the unknown variables is the total number N of all possible sensor arrangement positions of the bridge deck is 70); the genetic algorithm adopts a real number coding mode, the dimension of an unknown vector is 70, the unknown vector corresponds to all possible sensor layout positions, and upper and lower constraint boundaries of 0 and 1 are considered respectively. On the basis, the optimization problem of sensor layout combination can be solved by only applying the optimized variable translation interface of the target function of the extended standard genetic algorithm, and the calculation result of the optimization process is shown in fig. 4, wherein the number of sensors which can be laid is gradually increased from 1 to 6 calculation conditions, which correspond to (a) in fig. 4 to (f) in fig. 4 respectively, are considered. Without loss of generality, the optimized layout condition of 3 sensors, namely (c) in fig. 4 is used for example, the upper graph shows a convergence curve of the genetic algorithm applied to the method along with the genetic generation number, and it can be seen that the whole resolving process is converged very quickly, the objective function value is reduced quickly and is converged to the global optimum value of-38.0785 quickly; the lower graph shows the intermediate results of sensor layout corresponding to each genetic generation number, and the vertical axis represents all possible sensor layout positions, namely three gray rectangles on the same vertical line correspond to one sensor layout scheme, so that it is obvious that as the evolution process of the algorithm advances, the sensor layout rapidly reaches stability and points to the global optimal configuration scheme {26,35,42}, and the corresponding target function value is-38.0785.

By referring to the precise combination results given by the exhaustive method, table 1 compares the accuracy and efficiency of solving the problem of optimal configuration of the sensor based on the genetic algorithm extended by the present invention. It is clear from the table that for the 6 calculation conditions considered, the global optimum combination is found accurately by applying the present invention. It should be noted that, in the case of a small number of sensors that can be arranged, the calculation time of the exhaustive method is a certain advantage, but as the number of sensors increases gradually, the total number of all possible combinations increases in a geometric progression, and the calculation time of the exhaustive method also increases dramatically, in contrast to the genetic algorithm applying the present invention, which has a small calculation time and is not significantly affected by the increase of the number of sensors. Particularly, when 6 sensors are optimally arranged at 70 possible positions, the total combination number is as much as 1.3 hundred million, the calculation time of the exhaustive method is nearly 2000 times of the time of the resolving technology, the memory of the computer is consumed by the exhaustive method to be huge, and with the further increase of the number of the sensors, the calculation time and the calculation hardware consumption of an ordinary personal computer are far from being satisfied. Aiming at solving the high-dimensional combination optimization problem, the constraint gradient-free optimization algorithm applying the expansion interface can be efficiently solved. It should be noted that, although the present embodiment is shown only by taking the conventional genetic algorithm as an example, the present invention can be generally applied to other gradient-free optimization algorithms with boundary constraint, such as a particle swarm algorithm, an ant colony algorithm, a simplex algorithm, and the like, without changing the original algorithm code, so that these algorithms can be used for effectively solving the high-dimensional combinatorial optimization problem in the optimized layout of the large bridge structure sensor.

TABLE 1 solving sensor optimal configuration problem by applying the genetic algorithm of the extended translation interface of the present invention

In the description of the present invention, "a plurality" means two or more unless otherwise specified; the terms "upper", "lower", "left", "right", "inner", "outer", "front", "rear", "head", "tail", and the like, indicate orientations or positional relationships based on the orientations or positional relationships shown in the drawings, are only for convenience in describing and simplifying the description, and do not indicate or imply that the device or element referred to must have a particular orientation, be constructed in a particular orientation, and be operated, and thus, should not be construed as limiting the invention. Furthermore, the terms "first," "second," "third," and the like are used for descriptive purposes only and are not to be construed as indicating or implying relative importance.

It should be noted that the embodiments of the present invention can be realized by hardware, software, or a combination of software and hardware. The hardware portion may be implemented using dedicated logic; the software portions may be stored in a memory and executed by a suitable instruction execution system, such as a microprocessor or specially designed hardware. Those skilled in the art will appreciate that the apparatus and methods described above may be implemented using computer executable instructions and/or embodied in processor control code, such code being provided on a carrier medium such as a disk, CD-or DVD-ROM, programmable memory such as read only memory (firmware), or a data carrier such as an optical or electronic signal carrier, for example. The apparatus and its modules of the present invention may be implemented by hardware circuits such as very large scale integrated circuits or gate arrays, semiconductors such as logic chips, transistors, or programmable hardware devices such as field programmable gate arrays, programmable logic devices, etc., or by software executed by various types of processors, or by a combination of hardware circuits and software, e.g., firmware.

The above description is only for the purpose of illustrating the present invention and the appended claims are not to be construed as limiting the scope of the invention, and any modifications, equivalents and improvements made by those skilled in the art within the spirit and principle of the present invention are intended to be covered by the present invention.

Claims

1. A method for optimizing the layout of a structural health monitoring sensor with constrained non-gradient general solution is characterized by comprising the following steps of:

constructing a target function for optimizing the layout of the sensors, expanding a vector translation interface for optimizing the target function, and obtaining a sensor layout optimization combination scheme by using constraint gradient-free general solution;

the extended objective function optimization vector translation interface comprises:

(1) will be provided withNPerforming descending arrangement on the dimension continuous real number optimization vectors; produced by extractionNA dimension ordering number vector; intercept theNTop of dimension ordering number vectorMA number element;

(2) independent generationNA dimension 0 element vector; by interceptionMGenerated by pairs of dimension-numbered vectorsNThe vector of dimension 0 elements is subjected to a transformation operation, i.e. theNDimensional 0 element vector neutralizationMSetting the element corresponding to the number value to be 1 and keeping the rest elements unchanged, and finishing the operationNDimension continuous vector directionNTranslation of a vector of elements type dimension 01;

(3) will obtainNThe dimension 01 type element vector is used as an indication vector of the sensor layout position, and the indication vector is used for evaluating a utility function value of a sensor corresponding to a layout scheme;

(4) further using the utility function value as an origin outside the interfaceNThe target function value actually corresponding to the dimensional continuous real number vector completes the utility evaluation process of the continuous optimization vector translation coding;

2. The method for optimizing the layout of structural health monitoring sensors by constraining gradient-free general solution according to claim 1, wherein in the first step, the determining the target structural sensor optimal layout criterion comprises:

analyzing a design drawing of a structure to be monitored and other related technical data, comprehensively considering the characteristics of a space structure system of the structure to be monitored, establishing a finite element numerical model of a target structure, and drawing out all possible sensor layout positions based on the model; determining key monitoring positions by combining the finite element model and the actual health monitoring requirements of the target structure, constructing a layout optimization criterion of the sensor system by taking the measurement information requirements quantitatively identified by the damage and degradation characterization parameters of the key monitoring positions as main targets in a probability theory and information theory framework, and enabling the sensor system to be laid according to the criterionMThe sensors with limited number can acquire relatively most abundant static and dynamic force measurement data information so as to meet the quantitative inversion information requirement of damage degradation parameters to the maximum extent.

3. The method of claim 2 for constrained non-gradient general solution structural health monitoring sensor placement optimization, the total number of all possible sensor placement locations totalingN。

4. The method for optimizing the layout of structural health monitoring sensors by constraining gradient-free general solution as claimed in claim 1, wherein in step one, the gradient-free optimization algorithm with boundary constraint comprises: the algorithm adopts continuous real number vector coding, and the upper and lower boundaries of each element in the unknown vector are respectively taken as 0 and 1.

5. The method for optimizing the layout of structural health monitoring sensors by constraining gradient-free general solution as claimed in claim 1, wherein in step two, the constructing an objective function of the optimized layout of sensors comprises:

the sensor configuration scheme is represented by a continuous real number optimization vector in a gradient-free optimization algorithm with boundary constraint, each sensor layout scheme quantitatively evaluates a utility function value according to a layout criterion, and a sensor layout target function is constructed.

6. The method for optimizing the layout of structural health monitoring sensors by constraining general gradient solution according to claim 5, wherein each element in the real number type optimization vector corresponds to a specific sensor layout position on the structure to be monitored; optimizing the vector Total dimensionNRepresenting the total number of all possible sensor arrangement positions proposed.

7. A program storage medium receiving user input, the stored computer program causing an electronic device to perform a method of structural health monitoring sensor layout optimization of a constrained gradient-free general solution as recited in any one of claims 1-6, the method comprising the steps of:

and secondly, constructing a target function for optimizing the layout of the sensors, expanding a vector translation interface for optimizing the target function, and obtaining a sensor layout optimization combination scheme by using constraint non-gradient general calculation.