DE102020108034B3 - Modular multilevel converter, method for operating modular multilevel converters and computer program - Google Patents

Abstract

The invention relates to a modular multilevel converter which emits an electrical output power that comprises an output voltage and an output current, the converter having a circuit arrangement for transmitting the output power, which a) has at least a first, a second and a third converter branch, wherebyib) the first converter branch is connected to a first potential, c) the second converter branch is connected to a second potential that differs from the first potential, and d) the third converter branch is connected to a third potential that is between the first and the second potential is, wherebyie) the first, the second and the third converter branch are connected to one another at an output connection of the circuit arrangement at which the output power is transmitted, wherebyif) the first and the second converter branch are each designed as multilevel converter branches, in each of which individual modules hint are connected in series.

Description

1. a) wenigstens einen ersten, einen zweiten und einen dritten Umrichterzweig aufweist, wobei
2. b) der erste Umrichterzweig an ein erstes Potential angeschlossen ist,
3. c) der zweite Umrichterzweig an ein zweites Potential angeschlossen ist, das sich von dem ersten Potential unterscheidet, und
4. d) der dritte Umrichterzweig an ein drittes Potential angeschlossen ist, das zwischen dem ersten und dem zweiten Potential liegt, wobei
5. e) der erste, der zweite und der dritte Umrichterzweig an einem Ausgangsanschluss der Schaltungsanordnung, an dem die Ausgangsleistung übertragen wird, miteinander verbunden sind, wobei
6. f) der erste und der zweite Umrichterzweig jeweils als Multilevel-Umrichterzweige ausgebildet sind, in denen jeweils Einzelmodule hintereinandergeschaltet sind.

The invention relates to a modular multilevel converter which emits an electrical output power which comprises an output voltage and an output current, the converter having a circuit arrangement for transmitting the output power

1. a) has at least a first, a second and a third converter branch, wherein
2. b) the first converter branch is connected to a first potential,
3. c) the second converter branch is connected to a second potential that differs from the first potential, and
4. d) the third converter branch is connected to a third potential which lies between the first and the second potential, wherein
5. e) the first, the second and the third converter branch are connected to one another at an output connection of the circuit arrangement at which the output power is transmitted, wherein
6. f) the first and second converter branches are each designed as multilevel converter branches in which individual modules are connected in series.

The third converter branch can be designed differently.

The invention also relates to a method for operating such a modular multilevel converter of the aforementioned type and a computer program for carrying out the method.

The procedure is based on the control procedure according to [1], in which the branch currents and energies are actively controlled. The control method presented can be easily applied to the MMC topology shown for the first time in [2]. This topology is derived from the conventional MMC topology by adding a branch with full bridge modules between the midpoint of the converter phase and the intermediate circuit neutral point. By means of the regulation proposed here, this additional middle MMC branch can be replaced by a bidirectional switch. It is thus possible to combine the advantages of conventional MMC topologies (voltage scalability, redundancy, small output voltage levels) and the output voltage curve of quasi-three-level operation, while at the same time simplifying the topology and significantly reducing the required module capacity. This promises improvements in terms of volume, power density, efficiency and costs.

Figure list

• : Two-point inverter for low-voltage applications on the left and medium-voltage applications on the right
• : Example of the pulse duration modulation of the target output voltage u _a * over an output period for two-point inverters, cf. [4]
• : three-phase T-type inverter [6]
• : MMC topology with three inverter phases
• : Half-bridge modules and full-bridge modules
• : Single-phase model of the MMC
• : Example of a multilevel output voltage [10]
• : Quasi-two-level operation of a single-phase MMC over a PWM period
• : Topology of the single-phase MMC for quasi-three-level operation [2]
• : Output voltage curve at quasi-three-level over two output periods [2]
• : Module capacitor voltages above: Two A; Middle: branch B; below: branch M [2]
• : Control structure for quasi-three-level operation
• : Single-phase three-level MMC
• : State machine current control
• : Quasi-three-level operation with the proposed control concept
• : Small voltage levels at the output terminals
• : Topology of the three-phase hybrid MMCs for quasi-three-level operation with two anti-parallel thyristors in the middle branch
• : Topology of the three-phase hybrid MMCs for quasi-three-level operation with two counter-connected IGBT or IGCT with anti-parallel diodes
• : Transition A → M with positive output current
• : Transition M → A with positive output current
• : Quasi-three-level operation of the new hybrid MMC topology

State of the art

This chapter presents the converter topologies and operating modes that are relevant for this patent application. First, two-point inverters, three-point inverters and MMCs in conventional operation and in quasi-two-level operation will be presented. The quasi-three-level operation according to [2] is then presented.

It is additionally applied to the WO 2019/238443 A1 pointed out, from which a modular multi-file converter with the features of the preamble of claim 1 emerges.

Two-point inverter

The two-point inverter is the most common inverter in the field of low-voltage applications.

The two-point inverter is in shown on the left and is characterized by the simple control and the simple structure. The two-point inverter has six switching elements 1 (here, for example: IGBT with anti-parallel diode), each of which must be able to block _{the input voltage U e.} The maximum reverse voltage of IGBT is currently 6.5 kV, which limits the voltage scalability. In order to overcome this, several switching elements 1 can be connected in series in order to minimize the voltage load on the individual switching element 1. This is in shown on the right. The challenge here is the even voltage distribution and simultaneous control of the switching elements 1. This requires special, highly developed gate units [3].

Redundancy can only be made possible with press-pack switches, as these switch to the defined short circuit in the event of a fault. The desired sinusoidal output voltage is achieved by means of pulse duration modulation (PWM) as a short-term average. Either the upper switch or the lower switch is switched on here. The result is as in Typical two-point output voltage curve shown for the target output voltage u _a *.

Due to this two-stage modulation, the output voltage and the output current are subject to relatively high harmonics, depending on the switching frequency, which can lead to additional losses in electrical machines. Since the medium-voltage semiconductor switching elements can only be switched with switching frequencies of a few hundred Hertz, the harmonics are relatively high.

Also due to the two-stage operation, the output voltage jump with each switching process is the same as the input voltage. In the case of long cables to the electrical machine, this voltage jump leads to wave reflection. As a result, the insulation of the machine is loaded with up to twice the input voltage, which significantly increases the demands on the machine's insulation systems [5]. These problems are even more important in medium voltage applications, in which the isolation systems for the machines are already very demanding.

Accordingly, this topology is not suitable for use with electrical machines in the medium-voltage range. The limiting factor here is not so much the harmonic current, which leads to somewhat increased losses, than the increased stress on the insulation caused by line reflections, which can lead to premature machine failures and is therefore unacceptable.

For this reason, an additional filter is often installed at the output of the two-level inverter, which makes the entire system larger and more expensive.

Three-point inverter

With three-point inverters, the output voltage can be _{modulated over three voltage levels (U e} / 2, 0 V, -U _e / 2), which improves the harmonic properties of the output signals and halves the output voltage jump compared to the two-point inverter . This reduces the load on the electrical machine's insulation system.

Topologies frequently mentioned in the literature for three-point inverters are NPC converters, flying capacitor converters and T-type converters [5]. The T-Type converter is described below as an example.

The structure of a T-type inverter is in shown. A capacitive voltage divider C ₁ , C _{2 forms} the neutral point (NP) to which a further branch with a bidirectional power electronic switch is connected. This increases the number of switching elements. During operation, the voltage balancing of the input capacitors must also be guaranteed, which increases the control effort [6].

To generate the three output voltages, either the upper switch, the lower switch or the middle switch is switched on, while the other switches block.

The upper and lower switches must be able to block the entire input voltage. The middle switches, on the other hand, must be able to block half the input voltage [6]. For voltage scalability, as with the two-point inverter, several switches can be connected in series, whereby the same challenges arise in terms of voltage distribution and simultaneous control.

Compared to the two-point inverter, the harmonic properties of the output signals are improved and the output voltage jump is halved. However, the number of power electronic switching elements increases and the task of neutral point balancing is added. When it comes to voltage scalability, the same challenges arise as with two-point inverters.

With three-point inverters, the harmonic currents can be reduced significantly compared to two-point inverters, and the voltage jumps at the inverter output are halved. With modern insulation systems, a medium-voltage drive can therefore be operated directly at the converter output without a filter [7]. However, for motors with classic insulation systems (i.e. also with machines in the field that are to be retrofitted with a converter, so-called retrofit application), filters are also required between the converter and the motor.

Harmonics are no longer limiting here, but the voltage jumps in combination with the motor insulation are.

Modular multilevel converter

The modular multilevel converter (MMC) [8] is in shown schematically.

An MMC consists of several inverter phases (typically three) that are connected to the input. Each of these converter phases contains two branches. The output system is connected between the branches of each inverter phase. The branches consist of several n _mod modules and a branch inductance L _z [9]. As shown in [5], the inductances can be coupled to one another.

The most frequently used modules are half-bridge modules (HB) and full-bridge modules (VB). These are in the shown.

Half-bridge modules consist of two switches and a capacitor. The module voltage U _mod can assume either 0 V or the module _{capacitor voltage U Cmod, depending on the switching state.}

Full bridge modules consist of two additional failures, whereby the module voltage can assume 0 V, the positive module _{capacitor voltage} u Cmod and the negative module _{capacitor voltage} (- u Cmod).

Will be like in If the modules are connected in series, high operating voltages can be achieved, while the power semiconductors in the modules only have to be designed for the module capacitor voltage. This enables simple voltage scaling.

Redundancy can be implemented comparatively easily by adding several modules in series and bridging the defective modules in the event of a fault.

The topology is therefore more complex compared to the two-level inverter and three-level inverter, and the control is also more demanding. It must be guaranteed that the module capacitor voltages are equalized (balanced) during operation, which is ensured in branch modulation processes.

To simplify the description, the relevant variables are marked on the single-phase model of the MMC. The upper branch is referred to below as branch A and the lower branch as branch B. The branch voltage is marked with u _z and the branch _{current with i z} . The output voltage of the converter is described with u _a and the output current with ia.

Conventional operation [10]

In conventional operation, the output voltage is finely modulated. The example in voltage curve shown at the output.

This finely graded modulation results in a lower harmonic load on the output voltage and the output current and an additional filter can be saved. This also reduces the voltage jump at the output terminals compared to the two-point inverter and three-point inverter, which leads to a minimized voltage stress on the insulation of the electrical machine.

During DC / AC operation, branch powers p _{z are} implemented, which oscillate with the single or double output angular frequency. This generally leads to relatively large branch energy fluctuations e _z , which have to be buffered in the module capacitors. The ratio H of the energies stored in the modules to the apparent power of the MMC can be given as a measure for this. $$\text{H =}\frac{{\text{E.}}_{{\text{C.}}_{\text{mod}}}}{\text{S.}}$$

For MMCs in conventional operation, this is around 55 mJ / VA [5]. Accordingly, the module capacitors for conventional MMC operation must be dimensioned comparatively very large. As a result, the modular capacitors represent a large part of the volume, weight and cost of the converter.

The branch energy fluctuation and the associated module capacitor capacity to be installed is inversely proportional to the output frequency. So that the voltages of the capacitors do not rise excessively, which could lead to the destruction of the power semiconductors in the modules, special operating modes must be used in conventional operation at low frequencies close to zero. A typical operating mode is the “low-frequency mode” by Korn et al., Which was presented in [11]. Although the energy fluctuation in the capacitors is significantly reduced by this operating mode, it leads, as [12] shows, to a significantly increased current load on the power semiconductors when high currents are required at low frequencies. This has a negative effect on dimensioning and efficiency.

Quasi-two-point PWM operation [1,5,13-15]

In the quasi-two-point PWM operation (quasi 2-level operation, Q2L) according to [13], the conventional MMC is switched off operated similar to a two-point inverter. The output voltage becomes also modulated by the voltage levels U _e / 2 and -U _e / 2. For this purpose, one branch provides approximately the value of the full input voltage (active branch) and the other branch 0 V (passive branch).

The basic idea is that the majority of the output current should always be routed via the passive branch so that the branch power is as small as possible and accordingly low values are required for module capacities.

The distribution of the output current between the branches is determined by the phase cross current i _Q (see Sect. ) is set, which flows through both branches as a superposition current. In the quasi-two-level PWM mode, the phase cross-current is actively controlled.

For this purpose, the nominal branch voltage is generated in the active branch using high-frequency modulation. If the active and passive branch should change, the branch voltages are set in [1] so that the phase cross current changes as quickly as possible. For this purpose either both branches generate 0V or the maximum branch voltage level. This is referred to as a transient transition in the following.

During the transient transitions, however, high branch voltages overlap with high branch currents. This leads to power peaks that have to be buffered by the capacitors and remove the capacitor voltages from the desired setpoint values.

In order to compensate for this, a small current is to be impressed in the active branch using the phase transverse current in order to compensate for the branch energy swing (compensation current). This represents the second basic idea of the quasi-two-level PWM operation. The level of the current for the compensation is set via a branch energy regulator or is calculated predictively.

Thus, the branch energies can be regulated within each PWM period and the branch energy fluctuation is independent of the output frequency. The module capacity to be installed is therefore significantly reduced (by more than an order of magnitude compared to conventional operation).

A waiting time is implemented between the switching on and off of the modules within a branch in order to continue to guarantee the low output voltage levels. In this way the rate of rise of the output voltage can be limited and the overvoltages due to long machine cables can be reduced significantly [5].

The modular structure, the easy-to-implement redundancy, the voltage scalability and the low voltage jumps in the output voltage are retained in the quasi-two-level PWM operation. In addition, this operation enables the module capacity to be significantly reduced, which leads to a significant reduction in costs and an increase in power density.

The disadvantage compared to conventional operation is the worsened harmonic spectrum of the output voltage, which, however, can be accepted in many machines as long as the frequency of the PWM is not too low.

It should be noted here that in addition to the quasi-two-level PWM operation presented here, there are also quasi-two-level operations that do not actively regulate the currents and energies, e.g. [16-20]. This usually leads to additional disadvantages such as higher peak currents in the branches, operating behavior strongly influenced by parasitic parameters (e.g. losses and stray inductances of the connection technology) and increased module capacities.

Modular multilevel converter for quasi-three-level operation [2]

The quasi-three-level operation for MMCs was presented for the first time in 2018 in [2] and takes up the idea of the quasi-two-level operation. The output voltage is to be _{modulated by the voltage levels U e} / 2, 0 V and -U _e / 2 so that the output spectra of the converter can be improved with the same switching frequency. Since both branches have to provide half the input voltage during the output voltage level 0 V, an additional branch is necessary so that there is the possibility of routing the output current via a passive branch.

The presented topology consists of half-bridge modules in branches A and B of a converter phase and full-bridge modules in the additional middle branch M (here: clamped arm). The topology is saved as a “Quasi Three-level Hybrid Modular Multilevel Converter” introduced, whereby “Hybrid” can be traced back to the use of half-bridge and full-bridge modules. With this topology, the number of full bridge modules to be installed in the middle branch M corresponds to half the number of half-bridge modules to be installed in the upper and lower branches A, B.

In order to set the correct output voltage, the branch voltages are switched directly from one static state to the next. Accordingly, the branch currents are not actively regulated, but oscillate with the resonant circuit behavior of connected module capacitors and branch inductance (similar to the one described at the end of Section 1.3.2).

The output voltage curve has three static output voltage levels, as in can be seen.

A waiting time between the connection or bridging of the individual modules is also implemented for the edges in order to ensure the low output voltage levels.

Since no compensation currents are implemented, the energy swings are not regulated due to the transient transitions, which means that the module capacitor voltage fluctuation range is smaller than in conventional operation of the MMC, but still depends on the output frequency. This can also be seen in the following figure.

It should also be noted that the branch currents have not been shown in [2] and their peak values are to be expected to be very high.

invention

Compared to the prior art, the improvements and advantages mentioned in the introduction can be realized by the features specified in the independent claims of this patent application.

According to one embodiment of the invention, the modular multilevel converter is simplified with regard to its structure in that the third converter branch consists of a bidirectional switch and one or no inductance. Accordingly, the hardware outlay for a third multilevel converter branch is saved and only a comparatively simple circuit is provided as the third converter branch. In particular, compared to a third full bridge branch, there is a considerable reduction in effort.

In addition to the series connection of the individual modules, the first and the second converter branch can also have other components, such as at least one inductance connected in series. The individual modules of a converter branch can be individual modules of the same type or different individual modules.

According to an advantageous embodiment of the invention, it is provided that the bidirectional switch has an arrangement of power semiconductor switching elements that can symmetrically absorb reverse voltage and symmetrically conduct current. In this way, the bidirectional switch can be provided inexpensively with a simple structure. In addition, the electrical requirements can be reliably met. The arrangement of power semiconductor switching elements can, for example, have anti-parallel connected thyristors, oppositely connected series-connected IGBT or IGCT switching elements with anti-parallel diodes or a combination thereof, in particular a series connection of such elements.

The modular multilevel converter, hereinafter also referred to as converter for short, can be connected to three mutually independent input potentials (first, second and third input potential). The third input potential can also be formed from the first and the second input potential, e.g. by means of a capacitive voltage divider. According to an advantageous embodiment of the invention it is provided that the third potential is provided by a circuit arrangement that has a first capacitor, which is connected to the first potential with a first terminal and a second terminal to the third potential, and a second capacitor, which is connected with a first connection to the second potential and a second connection to the third potential.

According to an advantageous embodiment of the invention, it is provided that each individual module has a circuit arrangement connected to module connections of the individual module and made up of at least two internal Having switching elements and a module capacitor. The individual modules can be designed as half-bridge modules or as full-bridge modules. In the case of half-bridge modules, the module output voltage can assume the value 0 volts or positive module capacitor voltage. In the case of full-bridge modules, the module output voltage can assume the value 0 volts or positive or negative module capacitor voltage.

According to an advantageous embodiment of the invention, it is provided that the switching elements of an individual module are arranged such that either the positive voltage of the module capacitor can be applied to the module connections or the module connections can be connected directly to one another.

According to an advantageous embodiment of the invention, it is provided that the switching elements are arranged so that either the positive voltage of the module capacitor or the negative voltage of the module capacitor can be applied to the module terminals or the module terminals can be connected directly to one another.

• h) die Stromstärke des Ausgangsstromes wird durch ein Modulationsverfahren der Ausgangsspannung geregelt, wobei einer der Umrichterzweige als passiver Umrichterzweig betrieben ist, bei dem jeweils zwischen den Modulklemmen jedes Einzelmoduls, wenn es ein Multilevel-Umrichterzweig ist, oder am bidirektionalen Schalter keine Spannung anliegt und durch den der Ausgangsstrom fließt,
• i) in allen Multilevel-Umrichterzweigen sowie im geschlossenem bidirektionalem Schalter werden Ströme geregelt, die zur Kompensation von Energieschwankungen der Modulkondensatoren in den Multilevel-Umrichterzweigen dienen, durch ein Modulationsverfahren des Multilevel-Umrichterzweigs oder der Multilevel-Umrichterzweige, durch die der Ausgangsstrom nicht fließt und die somit aktive Umrichterzweige sind.

Another embodiment of the invention relates to a method for operating a modular multilevel converter, which is designed as mentioned above. The third converter branch can either be designed as a multilevel converter branch with several individual modules connected in series or consist of a bidirectional switch and one or no inductance. The operation of the converter is controlled, which has at least one control strategy A with the following features:

• h) the amperage of the output current is regulated by a modulation process of the output voltage, whereby one of the converter branches is operated as a passive converter branch, in which no voltage is applied between the module terminals of each individual module, if it is a multilevel converter branch, or at the bidirectional switch and through which the output current flows,
• i) In all multilevel converter branches as well as in the closed bidirectional switch, currents are regulated, which serve to compensate for energy fluctuations in the module capacitors in the multilevel converter branches, by a modulation method of the multilevel converter branch or the multilevel converter branches, through which the output current does not flow and which are therefore active converter branches.

For the regulation of the branch currents, these are measured according to an advantageous embodiment of the invention and compared with target values.

According to an advantageous embodiment of the invention it is provided that the modulation method of the active converter branch or the active converter branches is carried out with a frequency which is higher than a pulse duration modulation frequency with which the output voltage is synthesized. With such a relatively high-frequency modulation frequency in the active converter branch or the active converter branches, the desired regulation can be carried out particularly precisely. Assuming that the output voltage is an alternating voltage with a basic frequency, for example 50 Hz, the output voltage in quasi-three-level mode, in which the converter is operated, is carried out with a pulse duration modulation frequency that is at least twice as high as the Fundamental frequency is. The frequency with which the modulation method of the active converter branch or the active converter branches is carried out can, for example, be at least twice as high as the pulse duration modulation frequency with which the output voltage is synthesized.

According to an advantageous embodiment of the invention it is provided that the control of the operation of the converter additionally has at least one control strategy B in which the fastest possible change in the currents in the converter branches is caused by the individual modules in the multilevel converter branches.

According to an advantageous embodiment of the invention, it is provided that the control of the operation of the converter alternates between the at least two control strategies A and B.

According to an advantageous embodiment of the invention, it is provided that a change is made from control strategy A to control strategy B when the modulation of the output voltage specifies a change in the passive converter branch.

According to an advantageous embodiment of the invention, it is provided that a change is made from control strategy B to control strategy A when the output current flows through the passive converter branch specified by the output modulation. In this way, the setpoint values of the regulation can be achieved particularly well.

According to an advantageous embodiment of the invention it is provided that the individual modules have controllable switches, a waiting time between the change of the switch position of the switches in the multilevel converter branches. Accordingly, the individual modules are switched with a time offset from one another. In this way, a steep, trapezoidal or stepped profile of the output voltage can be generated. In this way, the rate of rise of the output voltage can be limited and the overvoltages due to long machine cables can be significantly reduced.

According to an advantageous embodiment of the invention, it is provided that in the event that the converter branch that has the bidirectional switch is to change from the active to the passive converter branch, the multilevel converter branches change the voltage so that a voltage of approximately 0 Volt is applied and then the bidirectional switch is closed before control strategy B is applied. In this way, in particular the variant of the converter in which one converter branch has the bidirectional switch is switched in such a way that undesired voltage jumps in the output voltage are avoided or at least reduced.

According to an advantageous embodiment of the invention, it is provided that in the event that the converter branch, which has the bidirectional switch, is to change from the passive to the active converter branch and the current through this converter branch is close to 0 amperes, the multilevel converter branches adjust the voltage as follows change so that the voltage at the bidirectional switch differs by a maximum of a module capacitor voltage of 0 volts if it were open, and the bidirectional switch then opens at a current of approximately 0 amperes before control strategy A is applied. In this way, the converter branch, which has the bidirectional switch, can be switched particularly gently. In particular, voltage peaks due to inductive current changes can be avoided.

The aforementioned method can advantageously be carried out by a computer program in that the computer program is carried out on a computer. For example, it can be a computer of a control device of the converter.

Accordingly, the invention also relates to a modular multilevel converter of the type mentioned at the outset, in which the third converter branch is designed either as a multilevel converter branch with several individual modules connected in series or consists of a bidirectional switch and one or no inductance. In addition, the circuit arrangement has at least one control device which is set up to control the power semiconductors of the converter branches. The control device is set up to carry out a method of the aforementioned type, for example by executing a computer program on a computer of the control device.

As already mentioned above, the bidirectional switch can have an arrangement of power semiconductor switching elements that can symmetrically absorb reverse voltage and symmetrically conduct current.

According to an advantageous embodiment of the invention it is provided that the third potential is provided by a circuit arrangement that has a first capacitor, which is connected to the first potential with a first terminal and a second terminal to the third potential, and a second capacitor, which is connected with a first connection to the second potential and a second connection to the third potential.

According to an advantageous embodiment of the invention, it is provided that the third converter branch in the form of a multilevel converter branch has individual modules which have a circuit arrangement connected to module connections of the individual module and made up of at least four internal switching elements and a module capacitor. The individual modules can be designed as full bridge modules.

According to an advantageous embodiment of the invention, it is provided that the switching elements are arranged so that either the positive voltage of the module capacitor or the negative voltage of the module capacitor can be applied to the module terminals or the module terminals can be connected directly to one another.

In the context of the present invention, the indefinite term “a” is not to be understood as a numerical word. If, for example, a component is mentioned, this is to be interpreted in the sense of “at least one component”. As far as angles are given in degrees, these refer to a circle of 360 degrees (360 °). As far as a computer is mentioned, it can be set up to execute a computer program, for example in the sense of software. The computer can be designed as a commercially available computer, for example as a PC, laptop, notebook, tablet or smartphone, or as a microprocessor, microcontroller or FPGA, or as a combination of such elements. As far as regulation is mentioned, regulation differs from open-loop control in that regulation has a feedback or feedback of measured or internal values with which the generated output values of the regulation are in turn influenced in the sense of a closed control loop. In the case of a control, a variable is simply controlled without such feedback or feedback.

Embodiments

In the following, possible exemplary embodiments of the new control method for the quasi-three-level operation of an MMC are presented. It can be applied to the topology shown in [2] with certain advantages. Building on this, it is shown how the full bridge modules from branch M can be replaced by a bidirectional switch with the help of pilot controls, which is an additional advantageous innovation.

Embodiments for control methods

In contrast to the control method from [2], it is proposed to control all energies and all currents of the converter in a similar way to the quasi-two-level PWM operation [1]. On the one hand, this can reduce the peak values of the branch currents and, on the other hand, lead to smaller module capacitors, which significantly reduces the load and costs of the modules.

A cascaded control structure is exemplified for the control suggested. The signals and functions marked in gray are not required for the new topologies presented in Section 2.2, but rather when branch M is designed as a multilevel converter branch with several individual modules connected in series.

In principle, the output current i _{a is} regulated via the modulated output voltage. For this purpose, a higher-level regulation, which in is not shown, the desired degree of modulation δ (setpoint value of the output voltage) is specified. This setpoint is realized on the basis of three states (state A: output voltage u _a ≈ Ue / 2, state B: output voltage u _a ≈ -Ue / 2, state M: output voltage u _a ≈ 0). The sequence and duration of the states can be determined with the usual modulation methods for conventional three-level inverters (carrier-based methods, space vector modulation, optimized pulse patterns, etc.). To illustrate the control by way of example, a carrier-based method with the carrier signals c ₁ and c _{2 is} shown.

As with quasi-two-level PWM operation, the output current should mostly flow through the branch whose voltage is 0 V (passive branch), so that the branch power is as small as possible in this branch and the module capacitor voltages change only slightly.

The branch currents are regulated via two circular currents. Circulating _current i k1 is a superposition current that flows through branches A and M. Circulating _current i k2 is a superposition current that flows through branches B and M. This is also in to recognize.

The current control is implemented in a state machine that differentiates between the static and the transient states. This state machine is exemplified in shown. The shows the respective transitions between states A, B and M with the corresponding transients.

In the static states, the circulating currents are regulated with a high-frequency modulation of the nominal branch voltages in the active branches.

1. 1. Simultane Transienten: Beide Kreisströme ändern sich gleichzeitig jedoch unterschiedlich stark.
2. 2. Sequentielle Transienten: Beide Kreisströme ändern sich maximal schnell jedoch zeitlich versetzt.

If the states A, B or M change, the branch voltages are set in such a way that the circulating currents reach their new setpoint as quickly as possible, suitable for the coming state. As a Circulating current has to change by half the other circulating current, the invention includes two possible variants:

1. 1. Simultaneous transients: Both circular currents change at the same time, however, to different degrees.
2. 2. Sequential transients: Both circular currents change as quickly as possible, but staggered in time.

Table 2-1 and Table 2-2 are given to describe these transient transitions. When the states A ↔ M transition, branch A or branch M set their maximum or minimum voltage level depending on which circulating current should decrease and which should increase. When the states B ↔ M transition, branch B or M set their maximum or minimum voltage level depending on which circulating current is to decrease and which is to increase (see Table 2-1 and Table 2-2).

In the case of simultaneous transients, the branch that does not switch between the active and passive branch provides half the input voltage (see Table 2-1). With sequential transients, this branch changes between the input voltage and 0 V after half the transient period (see Table 2-2). Table 2- 1: Simultaneous transients crossing Change i _k1 Change i _k2 U _zA U _e.g. U _zM A↔M Sink Climb U _e U _e / 2 -U _e / 2 Climb Sink 0 V U _e / 2 U _e / 2 B↔M Climb Sink U _e / 2 U _e U _e / 2 Sink Climb U _e / 2 0 V -U _e / 2 Table: 2-2; Sequential transients crossing Change i _k1 Change i _k2 U _zA U _e.g. U _zM A↔M Sink Climb U _e U _e 0 V -U _e / 2 Climb Sink 0 V U _e 0 V U _e / 2 B↔M Climb Sink U _e 0 V U _e U _e / 2 Sink Climb U _e 0 V 0 V -U _e / 2

In these transitions, high branch voltage and high branch currents are active at the same time and thus cause branch power peaks, similar to the quasi-two-level PWM operation of MMCs. In order to regulate the energy swings caused by these commutation processes, compensation currents are impressed on the active branches with the aid of the circulating currents. These can be calculated by an energy regulator or predictively.

This reduces the branch energy fluctuation independently of the output frequency and significantly. Thus, the module capacities can be dimensioned significantly smaller.

The modules of a branch are balanced with the help of the module balancing units. The waiting time between connecting and bridging the modules is also implemented here. Thus, the branch voltages change within a short time by a maximum of the magnitude of a module capacitor voltage. As a result, the output voltage jump is also reduced.

The control can be applied to the topology shown in [2]. In this case, the compensation currents i _{GR, comp} i _{example, comp} and i _zM.komp introduced to compensate the module capacitor voltages in branch A, B and M.

$$H_{\lfloor 2\rfloor }=\frac{15\cdot 3\cdot 3\frac{1}{2}\cdot 2\text{mF}\cdot {\left(1750\text{V}\right)}^2}{\frac{3}{2}\cdot {\left(800\mathrm{\Lambda }\right)}^2\cdot \sqrt{{\left(5\mathrm{\Omega }\right)}^2-{\left(2\mathrm{\pi }\cdot 5\text{ Hz}\cdot \mathrm{33,3}\text{ mH}\right)}^2}}=\mathrm{29,36}\frac{\text{mJ}}{\text{VA}}$$

$$H_{Sim}=\frac{15\cdot 3\cdot \frac{1}{2}\cdot \mathrm{0,2}\text{ mF}\cdot {\left(680\text{ V}\right)}^2}{\mathrm{713,6}\text{ kVA}}=\mathrm{2,92}\frac{\text{m]}}{\text{VA}}$$

The shows the quasi-three-stage output voltage and the resulting sinusoidal output current over an output period in the upper diagram. Compared to it can be seen that the module capacitor voltage fluctuations are independent of the output frequency. If you compare the ratio of the energies stored in the modules to the apparent power of the converter from [2] (H _[2] ) to that from the simulations with the proposed control (H _Sim ), the strong reduction in module capacity by a factor of 10 becomes clear. $$H_{\lfloor 2\rfloor }=\frac{\mathrm{15th}\cdot 3\cdot \frac{1}{2}{\mathrm{C.}}_{mOd}\cdot u_{{\mathrm{C.}}_{\mathrm{<figure-callout\; id="2"\; label="m\; O\; d"\; filenames="DE102020108034B3\_0005.png,DE102020108034B3\_0006.png"\; state="\{\{state\}\}">m</figure-callout>}Od}}^2}{\frac{3}{2}\cdot {\text{i}}_a^2\cdot \sqrt{{\mathrm{R.}}^2+{\left(j\omega \mathrm{L.}\right)}^2}}$$

$$H_{\lfloor 2\rfloor }=\frac{\mathrm{15th}\cdot 3\cdot 3\frac{1}{2}\cdot 2\text{mF}\cdot {\left(1750\text{V}\right)}^2}{\frac{3}{2}\cdot {\left(800\mathrm{\Lambda }\right)}^2\cdot \sqrt{{\left(5\mathrm{\Omega }\right)}^2-{\left(2\mathrm{\pi }\cdot 5\text{Hz}\cdot 33.3\text{mH}\right)}^2}}=29.36\frac{\text{mJ}}{\text{VA}}$$

$$H_{\mathrm{S.}im}=\frac{\mathrm{15th}\cdot 3\cdot \frac{1}{2}\cdot 0.2\text{mF}\cdot {\left(680\text{V}\right)}^2}{713.6\text{kVA}}=2.92\frac{\text{m]}}{\text{VA}}$$

The detail section of the output voltage also shows that the step-shaped change in the branch voltages also ensures the small output voltage steps.

Exemplary embodiments of the new hybrid MMC topology for quasi-three-level operation and the precontrol for this

With a pre-control presented here and the regulation from the last chapter, it is possible to replace the middle full bridge branch with a simple power electronic bidirectional switch. The effort for branch M is greatly reduced as a result.

• 1) Abschalten des induktiven Zweigstromes i_zM: Wird der Schalter geöffnet, muss der Strom im mittleren Zweig 0 A betragen, da sonst die induktive Stromänderung eine Spannungsspitze hervorrufen würde, was den Schalter beschädigen könnte.
• 2) Kleine Spannungsstufen an den Ausgangsklemmen: Ist der Schalter im mittleren Zweig zugeschaltet, entspricht die Ausgangsspannung 0 V. Ist der Schalter geöffnet, entspricht die Ausgangsspannung der halben Differenz der Zweigspannungen B und A. Dies würde ohne zusätzliche Vorsteuerung beim Abschalten und Zuschalten des mittleren Schalters zu erhöhten Spannungssprüngen an den Ausgangsklemmen führen.

If the full bridges in branch M are replaced by a bidirectional switch, the compensation of the middle branch M becomes superfluous and the energy regulation is simplified. If the switch in branch M is open, the branch currents are controlled via the phase cross-current similar to the quasi-two-level PWM operation. If the switch is closed, the branch currents are regulated via the circulating currents in and in a manner similar to the quasi-three-level operation described above. If branch M is to consist of a switching element, however, two major new challenges arise:

• 1) Switching off the inductive branch _current i zM: If the switch is opened, the current in the middle branch must be 0 A, otherwise the inductive current change would cause a voltage spike, which could damage the switch.
• 2) Small voltage levels at the output terminals: If the switch in the middle branch is switched on, the output voltage corresponds to 0 V. If the switch is open, the output voltage corresponds to half the difference between branch voltages B and A. This would be done without additional pre-control when switching off and on the middle branch Switch lead to increased voltage jumps at the output terminals.

In order to meet the first requirement, power _{semiconductor switches} are selected, for example, which automatically begin to block when the current i zM crosses zero. The following two possibilities are presented for this. shows variant a of the hybrid MMC topology with two anti-parallel thyristors (Smpos and Smneg) per phase. shows variant b of the hybrid MMC topology with two counter-connected IGBT or IGCT with anti-parallel diodes (Smpos and Smneg) per phase. For a positive current i _zM , Smpos conducts. For a negative current i _zM , Smneg conducts. The midpoint voltage is set via the capacitive voltage divider Ce ₁ and Ce ₂ .

It should be noted that the middle branches can be implemented both without (as shown here) and with inductances. These variants are also part of the invention.

In order to cope with the second requirement, a precontrol is implemented in the control, for example. This is illustrated using the example of the simultaneous transients during the transition of the states A↔M ( ) and M → A ( ) for a positive output current. For the transition of the states B → M and M → B, the provisions for the branch voltages U _zA and U _{apply, for example} vice versa.

Transition of the states A → M ( ):

• (1) State A: Most of the output current flows through branch A. The compensation current i, _{for example} , _comp is used to compensate for the difference in branch energy in branch B.
• (2) The output voltage is specifically driven to 0 V. For this purpose, branch voltages in branches A and B must be set to U _e / 2. The switch Sm is open.
• (3) The switch S _m is closed. The branch voltages are then changed according to the regulations of the simultaneous or sequential transients in order to lead the branch currents to their setpoints.
• (4) State M: The switch S _m is closed. Most of the output current flows through branch M. The compensation _currents i zA, _komp and i, _{for example} , _komp are used to compensate for the branch energy difference in branch A and B.

• (1) Zustand M: Der Schalter S_m ist geschlossen. Der Ausgangsstrom fließt größtenteils über den Zweig M. Die Kompensationsströme i_zA,_komp und i_zB,_komp werden zum Ausregeln der Zweigenergiedifferenz im Zweig A und B genutzt.
• (2) Der Schalter S_m ist geschlossen und die Zweigspannungen in den Zweigen A und B werden entsprechend der Vorschriften für die simultanen oder sequentiellen Transienten geändert.
• (3) Der Strom i_zM ist nahe 0 A. Die Zweigspannungen in den Zweigen A und B stellen Ue/2 ± u_Cmod. Somit ändern sich die Zweigströme nur noch langsam. Wenn i_zM = 0 A erreicht und somit der Thyristor (Variante a) bzw. die in Reihe geschaltete Diode (Variante b) sperrt, ändert sich die Ausgangsspannung maximal um eine Modulspannungsstufe. Sind die Zweigströme in den Zweigen A und B noch relativ weit von ihren Sollwerten entfernt, kann anschließend wie bei den transienten Übergängen im Quasi-Zwei-Level-PWM-Betrieb die Überlappung der vollen Zweigspannung oder 0 V genutzt werden.
• (4) Zustand A: Da der Strom i_zM = 0 A entspricht, sperrt der Thyristor (Variante a) bzw. der IGBT kann nun geöffnet werden (Variante b). Der Ausgangsstrom fließt größtenteils über den Zweig A. Der Kompensationsstrom i_zB,_komp wird zum Ausregeln der Zweigenergiedifferenz im Zweig B genutzt.

Transition of the states M → A ( ):

• (1) State M: The switch S _m is closed. Most of the output current flows through branch M. The compensation _currents i zA, _komp and i, _{for example} , _komp are used to compensate for the branch energy difference in branch A and B.
• (2) The switch S _m is closed and the branch voltages in branches A and B are changed according to the regulations for the simultaneous or sequential transients.
• (3) The current i _zM is close to 0 A. The branch voltages in branches A and B represent Ue / 2 ± u _Cmod . Thus the branch currents change only slowly. When i _zM = 0 A and thus the thyristor (variant a) or the series-connected diode (variant b) blocks, the output voltage changes by a maximum of one module voltage level. If the branch currents in branches A and B are still relatively far from their setpoints, the overlap of the full branch voltage or 0 V can then be used, as with the transient transitions in quasi-two-level PWM operation.
• (4) State A: Since the current i _zM = 0 A, the thyristor blocks (variant a) or the IGBT can now be opened (variant b). Most of the output current flows through branch A. The compensation current i, _{for example} , _comp is used to compensate for the branch energy difference in branch B.

With this method, the switch in branch M is operated with zero-current switching and the switching voltage is either 0 V or a module capacitor voltage. Accordingly, the switching losses are extremely low and the switch in branch M is not heavily loaded.

The voltage scalability in the middle branch is retained, as the low switching voltages mean that the challenge of even voltage distribution when connecting and disconnecting several switching elements connected in series is eliminated.

Redundancy can be guaranteed with press-pack switches connected in series, as these switch to a defined short circuit in the event of a fault.

In the event that the entire middle branch M fails and goes into an open state, the topology can continue to be operated in quasi-two-level operation.

How out As can be seen, the power control further guarantees that the module capacitor voltage fluctuation ranges are low and independent of the output frequency.

Overview

• 1) Neuartige Regelung für die Drei-Level-MMC-Topologie aus [2]
• 2) Vereinfachung der Drei-Level-MMC-Topologie aus [2] und eine Erweiterung der Steuerung hierfür

Two innovations have been presented in this document:

• 1) New type of regulation for the three-level MMC topology from [2]
• 2) Simplification of the three-level MMC topology from [2] and an extension of the control for this

The first point was described in Chapter 2.1. The new type of regulation promises better control over the energies and currents of the converter. Accordingly, it is possible to reduce the peak currents in the branches and to make the capacitors smaller by a factor of 10, which has already been proven by simulations. The additional effort is a faster measurement of the branch currents and a more complicated implementation of the control, both of which do not, however, have a significant effect on the total costs.

• - Die Anzahl von Schaltelementen im mittleren Zweig M ist reduziert und deren Ansteuerung ist vereinfacht, da alle gleichzeitig und nicht versetzt angesteuert werden müssen.
• - Es werden keine Kondensatoren in mittleren Zweig M benötigt. Diese müssen nicht mehr balanciert werden und deren Spannungen müssen nicht mehr gemessen werden.
• - Die Schalter im mittleren Zweig M können für niedrige Durchlassverluste optimiert werden, da diese immer bei niedrigen Spannung sowie bei 0 A ein- und ausgeschaltet werden und dementsprechend auch keine Schaltverluste verursachen.

The second point was described in Section 2.2. If the three-level MMC topology from [2] is controlled with the control from Section 2.1, this can be simplified by replacing the middle full bridge branch with a bidirectional switch. This switch can be based on thyristors, IGBTs or IGCTs, for example. If this new topology is combined with the appropriate control, there are no disadvantages compared to the original topology from [2]. At the same time, expectations regarding costs and losses are improved. This is due to the following points:

• - The number of switching elements in the middle branch M is reduced and their control is simplified, since all must be controlled simultaneously and not offset.
• - No capacitors are required in the middle branch M. These no longer have to be balanced and their tensions no longer have to be measured.
• - The switches in the middle branch M can be optimized for low on-state losses, since they are always switched on and off at low voltage and 0 A and accordingly do not cause any switching losses.

literature

• [1] J. Kucka and A. Mertens, "Improved Current Control of a Quasi-Two-Level PWM-Operated Modular Multilevel Converter" in 2018 IEEE Energy Conversion Congress and Exposition (ECCE), Portland, OR, Sep. 2018 - Sep. 2018, pp. 2948-2955 .
• [2] X. Wei, S. Lu, X. Deng and S. Li, “A New Quasi Three-level Hybrid Modular Multilevel Converter” in 2018 IEEE International Power Electronics and Application Conference and Exposition (PEAC), Shenzhen, Nov. 2018 - Nov . 2018, pp. 1-6 .
• [3] M. Bruckmann, R. Sommer, M. Fasching and J. Sigg, "Series Connection of High Voltage IGBT Modules" in Conference record of the 1998 IEEE Industry Applications Conference .
• [4] D. Schröder, Power Electronic Circuits: Function, Design and Application, 3rd Edition Berlin: Springer, 2012.
• [5] J. Kucka, "Quasi-two-level PWM operation of modular multilevel converters: implementation, analysis, and application to medium-voltage drives".
• [6] S. Brüske, construction performance comparison and neutral point balancing for 3-level inverter topologies for use in electric vehicles. Dissertation. Kiel: Christian-Albrechts-Universität zu Kiel, 2016.
• [7] R. Sommer et al., “New medium voltage drive systems using three-level neutral point clamped inverter With high voltage IGBT” in Conference Record of the 1999 IEEE Industry “Applications Conference. Thirty-Forth IAS Annual Meeting (Cat.No.99CH363 70), Phoenix, AZ, USA, Oct. 1999, pp. 1513-1519.
• [8] A. Lesnicar and R. Marquardt, “An innovative modular multilevel converter topology suitable for a wide power range” in 2003 IEEE Bologna Power Tech Conference Proceedings, Bologna, Italy, Jun. 2003, pp. 272-277.
• [9] K. Sharifabadi, L. Hamefors, H. P. Nee, S. Norrga and R. Teodorescu, Design, control, and application of modular multilevel converters for HVDC transmission systems. Chichester West Sussex United Kingdom: IEEE Press Wiley, 2016.
• [10] D. Karwatzki, Analysis and Control of a Class of Modular Multi-Level Inverter Topologies. Dissertation. Hanover: Gottfried Wilhelm Leibniz University Hanover, 2017.
• [11] A. J. Korn, M. Winkelnkemper and P. Steimer, "Low output frequency operation of the Modular Multi-Level Converter" in 2010 IEEE Energy Conversion Congress and Exposition, Atlanta, GA, Sep. 2010 - Sep. 2010, pp. 3993-3997.
• [12] A. Antonopoulos et al., "Modular Multilevel Converter AC Motor Drives With Constant Torque From Zero to Nominal Speed", IEEE Trans. On Ind Applicat., Vol. 50, No. 3, pp. 1982-19931, 2014 .
• [13] A. Mertens and J. Kucka, “Quasi Two-Level. PWM Operation of an MMC Phase Leg With Reduced Module Capacitance ", IEEE Trans. Power Electron., Vol. 31, No. 10, pp. 6765-6769, 2016 .
• [14] J. Kucka and A. Mertens, "Common-Mode Voltage Injection Techniques for Quasi Two-Level PWM-Operated Modular Multilevel Converter3" in the 2018 International Power Electronics Conference (IPEC-Niigata 2018 -ECCE Asia), Niigata, May. 2018 - May. 2018, pp. 1904-1911 .
• [15] J. Kucka and A. Mertens, "Control for Quasi Two-Level PWM Operation of Modular Multilevel Converter" in 2016 IEEE 25th International Symposium on Industrial Electronics (ISIE), Santa Clara, CA, USA, Jun. 2016 - Jun. 2016, Pp. 448-453 .
• [16] MF Aiello and W. Barie, "High resonate pole frequency M2LC converter for medium voltage drive and energy sector applications" in 2015 IEEE Energy Conversion Congress and Exposition (ECCE): 20-24 Sept. 2015, Montreal, Quebec, Canada, Montreal, QC, Canada, 2015, pp. 3515-3521 .
• [17] MF Aiello, KS Berton and WG Barie, "M2LC System and Method for Controlling Same," US 9372 493 B2 .
• [18] F. Bertoldi, M. Pathmanathan, RS Kanchan, K. Spiliotis and J. Driesen, "Quasi-Two-Level Converter for overvoltage mitigation in medium voltage drives" in the 2018 International Power Electronics Conference (IPEC-Niigata 2018 -ECCE Asia), Niigata, May. 2018 - May. 2018, pp. 488-494 .
• [19] F. Bertoldi, M. Pathmanathan and R., S. Kanchan, "Quasi-two-level converter operation strategy for overvoltage mitigation in long cable applications" in the 2019 IEEE International Electric Machines & Drives Conference (IEMDC), San Diego, CA, USA, May. 2019 - May. 2019, pp. 1621-1627 .
• [20] MS Diab, GP Adam, BW Williams, AM Massoud and S. Ahmed, "Quasi two-level PWM operation of a nine-arm modular multilevel converter for six-phase medium-voltage motor drives" in 2018 IEEE Applied Power Electronics Conference and Exposition (APEC), San Antonio, TX, USA, Mar 2018 - Mar 2018, pp. 1641-1648.

Claims (15)

Modular multilevel converter which emits an electrical output power which comprises an output voltage and an output current, the converter having a circuit arrangement for transmitting the output power which a) at least a first, a second and a third converter branch (A. B, M) comprises, wherein b) the first converter branch (A) is connected to a first potential, c) the second converter branch (B) is connected to a second potential that differs from the first potential, and d) the third converter branch (M) is connected to a third potential, which lies between the first and the second potential, wherein e) the first, the second and the third converter branch (A. B, M) at an output connection of the circuit arrangement, at which the output power is transmitted, with one another are connected, wherein f) the first and the second converter branch (A, B) are each designed as multilevel converter branches, in each of which E individual modules (2) are connected in series, g) wherein the third converter branch (M) consists of a bidirectional switch (3) and one or no inductance, h) characterized in that the circuit arrangement has a control device which is set up to control the power semiconductors of the converter branches The control device is set up to carry out a method that controls the operation of the converter, which has at least one control strategy A with the following features: i) the current intensity of the output current is regulated by a modulation method of the output voltage, with one of the converter branches being passive Converter branch is operated, in which there is no voltage between the module terminals of each individual module, if it is a multilevel converter branch, or at the bidirectional switch (3) and through which the output current flows, j) in all multilevel converter branches as well as in the closed This bidirectional switch (3) regulates currents, which are used to compensate for energy fluctuations in the module capacitors in the multilevel converter branches, by means of a modulation method of the multilevel converter branch or the multilevel converter branches, through which the output current does not flow and which are therefore active converter branches. Modular multilevel converter according to Claim 1 , characterized in that the bidirectional switch (3) has an arrangement of power semiconductor switching elements which can symmetrically absorb reverse voltage and symmetrically conduct current. Modular multilevel converter according to one of the preceding claims, characterized in that the third potential is provided by a circuit arrangement which has a first capacitor (C ₁ ) connected with a first connection to the first potential and a second connection to the third potential is, and a second capacitor (C ₂ ), which is connected with a first terminal to the second potential and a second terminal to the third potential. Modular multilevel converter according to one of the preceding claims, characterized in that each individual module (2) has a circuit arrangement, connected to module connections of the individual module, comprising at least two internal switching elements (1) and a module capacitor. Modular multilevel converter according to Claim 4 , characterized in that the switching elements (1) of an individual module (2) are arranged so that either the positive voltage of the module capacitor can be applied to the module connections or the module connections can be connected directly to one another. A method for operating a modular multilevel converter which emits an electrical output power which comprises an output voltage and an output current, the converter having a circuit arrangement for transmitting the output power which a) at least a first, a second and a third converter branch (A. B, M), wherein b) the first converter branch (A) is connected to a first potential, c) the second converter branch (B) is connected to a second potential that differs from the first potential, and d) the third Converter branch (M) is connected to a third potential which lies between the first and second potential, where e) the first, second and third converter branches (A. B, M) are connected to an output terminal of the circuit arrangement at which the output power is transmitted, are connected to one another, wherein f) the first and second converter branches (A, B) are each embodied as multilevel converter branches ldet, in each of which individual modules (2) are connected in series, g) where the third converter branch (M) consists of a bidirectional switch (3) and one or no inductance, h) characterized in that the operation of the converter is regulated which has at least one control strategy A with the following features: i) the amperage of the output current is controlled by a modulation method of the output voltage, one of the converter branches being operated as a passive converter branch, with each individual module between the module terminals, if it is a multilevel converter branch or there is no voltage at the bidirectional switch (3) and through which the output current flows, j) in all multilevel converter branches as well as in the closed bidirectional switch (3) currents are regulated that compensate for energy fluctuations in the module capacitors in the multilevel converter branches serve by a modulation Process of the multilevel converter branch or the multilevel converter branches through which the output current does not flow and which are therefore active converter branches. Procedure according to Claim 6 , characterized in that the modulation method of the active converter branch or the active converter branches is carried out with a frequency which is higher than a pulse duration modulation frequency with which the output voltage is synthesized. Method according to one of the Claims 6 to 7th , characterized in that the control of the operation of the converter additionally has at least one control strategy B, in which the fastest possible change in the currents in the converter branches is caused by the individual modules (2) in the multilevel converter branches. Procedure according to Claim 8 , characterized in that the control of the operation of the converter between the at least two control strategies A and B alternates. Procedure according to Claim 9 , characterized in that a change is made from control strategy A to control strategy B when the modulation of the output voltage specifies a change in the passive converter branch. Procedure according to Claim 9 or 10 , characterized in that a change is made from control strategy B to control strategy A when the output current flows through the passive converter branch specified by the output modulation. Method according to one of the Claims 6 to 11 , characterized in that the individual modules (2) have controllable switches (1), a waiting time between the change of the switch position of the switches (1) in the multilevel converter branches. Method according to one of the Claims 6 to 12th , characterized in that in the event that the third converter branch is to switch from the active to the passive converter branch, the multilevel converter branches change the voltage so that a voltage of approximately 0 volts is applied to the bidirectional switch (3) and the bidirectional switch is then closed before control strategy B is applied. Method according to one of the Claims 6 to 13th , characterized in that in the event that the third converter branch is to change from the passive to the active converter branch and the current through this converter branch is close to 0 amperes, the multilevel converter branches change the voltage so that the voltage at the bidirectional switch ( 3) differs from 0 volts by a maximum of one module capacitor voltage if it were open, and the bidirectional switch (3) then opens at a current of approximately 0 amperes before control strategy A is applied. Computer program with program code means, set up to carry out a method according to one of the Claims 6 to 14th when the computer program is executed on a computer.

Images