DE102022109827A1 - Method and system for target current modulation of a modular multilevel converter - Google Patents

Abstract

The invention relates to a method for controlling a modular multilevel converter for a load, for example a traction motor of a vehicle. Based on setpoint voltage levels (105) provided by a controller for each switching cycle and a phase angle of the motor phase currents (106), an efficiency-optimized ideal current distribution (210) is determined on the one hand, and an offset current specification (220) that compensates for the state of charge of the individual modules is determined on the other hand, taking into account the state of charge of the individual modules. A target module current distribution (240) is determined by summing up the efficiency-optimized ideal current distribution (210) and the state of charge balancing offset current determination (220), and switching commands (109) for controlling the multilevel converter are determined on the basis of the target module current distribution (240). In addition, a system is claimed which allows the method to be carried out.

Description

The present invention relates to a method for controlling a modular multilevel converter for an AC load. Furthermore, a control system is claimed which allows the method to be carried out.

A modular multilevel converter or multilevel converter is an arrangement of several electrically connected modules, each of which has at least one energy store and several semiconductor switches for dynamic interconnection of the energy stores between the modules. Due to the interconnection, an alternating voltage can be generated from a direct voltage of the energy store, for example for operating an electrical machine. An example is that of R. Marquardt in the reference US2018/0109202 to name the modular multilevel converter described, which is also abbreviated as MMC or M2C.

A special further development of the MMC is the modular multilevel serial-parallel converter, e.g. disclosed in “Goetz, SM; Peterchev, A.V.; Weyh, T., "Modular Multilevel Converter With Series and Parallel Module Connectivity: Topology and Control," Power Electronics, IEEE Transactions on , vol.30, no.1, pp.203,215, 2015. doi: 10.1 109/TPEL.2014.2310225 . This is also referred to as MMSPC or M2SPC and, compared to the MMC, has the ability to also connect the energy stores of the modules in parallel. The M2SPC also generates voltage differences between at least two connection terminals, e.g. an energy network or a traction system of an electric car, by configuring the electrical interconnection of energy storage devices in the modules and by switching modulation between switching states. This makes it possible to form a large number of intermediate states in order to generate the desired output voltage.

It is problematic to find out from the large number of possible switching states those which represent an optimal solution with regard to predetermined criteria, such as, for example, power consumption that protects the energy store. In addition, this solution must be available for each switching cycle, i.e. in real time.

The pamphlet DE 10 2011 075 429 A1 describes a battery-electric motor vehicle drive system with several battery modules, an electric traction motor and a multilevel converter. The multilevel converter can be controlled in such a way that the battery modules are discharged evenly.

The pamphlet DE 10 2016 125 720 A1 discloses a modular energy storage converter system with a control device that is set up to control switches depending on a current state of charge of storage elements. The voltages or states of charge of the memory modules can thus be matched before the memory elements are connected in parallel.

In the US publication U.S. 2021/0155100 A1 describes a method for predicting phase shifts of currents on a primary and a secondary side of the converter. Losses in the power electronics can be minimized in this way.

Against this background, it is an object of the present invention to provide a method for controlling a modular multilevel converter, which generates a respective switching state for the modules of the modular multilevel converter in real time according to predetermined optimization criteria. Furthermore, a system is to be claimed which allows the method to be carried out.

• • von einer Regelung taktsynchron zu einem Schalttakt des modularen Multilevelumrichters die Sollstromvorgabe als Sollspannungslevel für die mindestens eine Phase und ein Phasenwinkel bereitgestellt werden und von der Steuerung verlustminimale Idealströme zu einer Idealstromaufteilung berechnet werden,
• • zusätzlich zu der Sollstromvorgabe ein Ladezustand für jedes Modul herangezogen wird und daraus für einen Ladezustandsausgleich innerhalb der Mehrzahl an Modulen lastbasierte Offsetströme zu einer Offsetstromaufteilung bzw. Offsetstromfestlegung berechnet werden,
• • Idealstromaufteilung und Offsetstromfestlegung aufsummiert und mittels vorgegebener Sättigungswerte begrenzt werden, wodurch die jeweiligen Zielströme in einer Zielstromaufteilung gebildet werden,

und wobei mittels einer Kostenfunktion diejenigen Schaltzustände der Mehrzahl an Modulen, welche die Zielstromaufteilung mit einer geringsten Abweichung realisieren, bestimmt werden und damit der jeweilige Schaltbefehl pro Schalttakt gebildet wird.A method for target current modulation of a modular multilevel converter is proposed in order to solve the above-mentioned task, in which the modular multilevel converter has a plurality of modules and is controlled by a controller. Each module of the plurality of modules has at least two controllable switches and at least one energy store. At least one strand is formed by lining up the modules. A respective phase for an electrical load is formed by the at least one strand. For each switching cycle, the controller provides a respective switching command, which includes all the switching states of the modules that are to be set, and transmits it to the plurality of modules. The respective switching command results in a respective current in the respective module of the plurality of modules, with the switching command being specified a target current to be implemented, with the respective target current being calculated by the controller based on a target current specification by

• • the target current specification as a target voltage level for at least one phase and a phase angle is provided by a controller synchronously with a switching cycle of the modular multilevel converter, and ideal currents with minimal losses are calculated by the controller for an ideal current distribution,
• • In addition to the target current specification, a state of charge is used for each module and from this for a state of charge equalization inner half the plurality of modules load-based offset currents are calculated to an offset current distribution or offset current definition,
• • Ideal current distribution and offset current definition are summed up and limited using specified saturation values, whereby the respective target currents are formed in a target current distribution,

and wherein those switching states of the plurality of modules which implement the target current distribution with the smallest deviation are determined by means of a cost function, and the respective switching command per switching cycle is thus formed.

The electrical load can be both an electrical machine, for example a traction machine of an electric vehicle, and a power grid, for example a three-phase grid. The current that occurs in the respective modules based on the respective switching status depends on the load.

If it is not possible for the controller to calculate a system enclosing the entire multilevel converter in real time, e.g. because a computing unit used has too little computing power, in an advantageous embodiment of the method according to the invention, before the controller is put into operation, at least one setpoint voltage level is determined for each combination and a phase angle, all possible switching states that implement the respective combination are determined, the ohmic losses to be expected are calculated and only switching states with minimal losses for the respective combination are stored in a characteristic map. In this case, the phase angle results from the measurements of the phase currents flowing in the electrical load, for example the traction machine, currently available to the controller. In order to save resources, it is sufficient to use symmetries to calculate the respective switching states for just one phase and an angular range from 0° to 90°. A complete course can then be obtained by vertical mirroring or horizontal mirroring with subsequent shifting by 180°. However, it is also conceivable to store over an entire angle range from 0° to 360°.

The map has a number of dimensions equal to the number of phases plus one dimension for the phase angle. With a three-phase current that is usually required for a traction machine, there are three phases and, accordingly, a four-dimensional characteristic diagram. The map can also be referred to as a lookup table, abbreviated LUT.

In one embodiment of the method according to the invention, the saturation values or limit values are limited towards numerically larger values by technical upper limits, as a result of which the respective module or the energy store of the respective module is not overloaded. Toward numerically smaller values, the saturation values are limited to the zero value, which means that the respective modules can only be relieved, since switching states that would lead to charging are penalized by the cost function. A limitation to a minimum current value i _min and a maximum current value i _max of the target currents i _target ensues on the basis of the current carrying capacity of the energy stores (charging/discharging).

In another embodiment of the method according to the invention, negative values are additionally permitted when calculating the respective target current if a recuperative state is present. As a result, the modules are charged in a targeted manner on the basis of an estimate of an electrical power output from the electrical machine. The estimate is based on a measured engine speed and the target current specification. Such a recuperation detection limits the target current distribution depending on a currently existing recuperative state.

In a further embodiment of the method according to the invention, the saturation values are dynamically adjusted during operation.

In yet another embodiment of the method according to the invention, a common mode voltage level, known as “common mode voltage” (CMV), is additionally applied when calculating the respective target current.

While in the prior art a CMV is often formed by a harmonic of an electrical fundamental (in phase with a phase voltage or a phase current) and is usually part of a modulation in order to increase a dynamic range, the method according to the invention does not know a CMV defined in this way and is not technically tied to a vibration form when impressing the DC voltage level. Rather, combinations with possible additional common-mode voltage levels are also taken into account in the calculation according to the invention of the ideal current distribution for a required target voltage level, without limiting the modulation range.

In an even further developed embodiment of the method according to the invention, all common-mode voltage levels that can be impressed are also taken into account in the cost function for all switching states.

In a continued embodiment of the method according to the invention, the target currents of the respective modules are specified directly.

In a continued further embodiment of the method according to the invention, all possible target currents are calculated for a switching clock which follows the current switching clock and also taken into account in the cost function. Further measures reducing the ohmic losses are thus advantageously possible. For example, a shift in a star point can reduce the modulation factor of a phase, which means that more modules can be connected in parallel. If a high current is currently flowing in this phase, this advantageously results in lower ohmic losses.

In a continued yet further embodiment of the method according to the invention, an average value of the associated module currents is calculated for equivalent switching states stored in the characteristics map with identical minimum losses, or an arbitrary selection of one of these equivalent switching states is made.

In a further developed embodiment of the method according to the invention, the target voltage level is formed by PI control and subsequent delta-sigma modulation on the basis of a target current specification, which is specified by a phase or mains current specification. The method according to the invention is also capable of being carried out with any regulator/modulator combination, as long as a setpoint voltage is present as the input variable.

The method according to the invention can be carried out on a modular multilevel converter with at least one phase. A modular multilevel serial-parallel converter, abbreviated MMSPC, with three strands and correspondingly three phases is considered below without restricting other possibilities of implementation on multilevel converters or multilevel converters of another type. A complete formal MMSPC description can be found in "Optimizing Utilization of an MMSPC with Model Predictive Control", T. Merz, C. Korte, E. Specht, M. Hiller, 2020, IEEE 21st Workshop on Control and Modeling for Power Electronics ( COMPEL): Aalborg, Denmark, November 9-12, 2020, At.Nr. 9265803, Institute of Electrical and Electronics Engineers (IEEE); doi:10.1 109/COMPEL49091.2020.9265803. With the help of this description, the resulting module currents for all switching options can be calculated during ongoing operation using system-inherent resistances, phase currents and voltage levels, or, if the computing power of the control is too low, they can also be calculated in advance. This is used to create the map.

Based on an exact network analysis of the modular multilevel converter under consideration, such as the MMSPC description cited above, and the characteristics maps that can be provided as a result with switching states that minimize losses, the method according to the invention causes an efficiency-optimized target current specification for each switching cycle of the multilevel converter with simultaneous active balancing of the state of charge of the energy storage devices in the modules. In addition, it is possible to use the imprint of DC voltage levels to carry out a phase-spanning state of charge equalization and to increase efficiency by reducing the modulation factor in phases with a high current amplitude together with the associated ohmic losses.

mit den Schaltbefehlen (s₁, s₂, s₃) der drei Phasen. Einen Einfluss auf Vorgaben von Phase-Phase-Spannungen am Ausgang des Umrichters hat das erfindungsgemäße Verfahren jedoch nicht. Es besteht somit kein Eingriff in die eigentliche Maschinenregelung.The method according to the invention is implemented, for example, on a signal processing level of the MMSPC and is carried out between the controller and a bus system for transmitting the respective switching command to the modules of the MMSPC. It is therefore located between a modulator (provision of target voltage level) and the MMSPC as an actuator. Discrete target voltage levels lvl _out,uvw per phase (U, V, W) based on the target currents or d/q target currents (i _d,sp , i _q,sp ), currently measured phase currents i _ph,uvw , a Engine speed n and current state of charge SoC of the individual modules required or provided by the control. Internally, the method according to the invention works with the respective target currents related to the individual modules as calculation variables. A switching vector is used as the output variable $\overrightarrow{s}$

with the switching commands (s ₁ , s ₂ , s ₃ ) of the three phases. However, the method according to the invention does not have any influence on specifications for phase-phase voltages at the output of the converter. There is therefore no intervention in the actual machine control.

For each switching cycle, an optimal ideal current distribution i _efficiency is determined based on the required discrete target voltage level lvl _out,uvw and the current composition of the phase currents i _ph,uvw that can be derived from the phase angle: $$i_{\text{efficiency}}={ƒ}_{i,\text{efficiency}}\left(lv{l}_{\text{out}\text{,and many more}},i_{\text{Ph}\text{,and many more}}\right)$$

aus Ersatzwiderständen in dem Strang mit jeweiliger Phase R_ph sowie den jeweiligen Phasenströmen z_ph,uvw. Bei Gleichung (2) gilt die Annahme, dass die einzelnen Energiespeicher einen identischen Ladezustand aufweisen und somit bei einem Parallelschalten keine Ausgleichströme fließen. Ausgleichströme belasten die Batterien zusätzlich und führen somit zu weiteren Verlusten.In this case, minimal ohmic losses are optimized both in the switches or semiconductor switches and in the energy stores. In this case, all degrees of freedom, in particular a zero voltage, are used. The ohmic losses P _v,R are recalculated $$P_{\text{v},\text{R}}=R_{\text{Ph}\text{,u}}{i}_{\text{Ph}\text{,u}}^2+R_{\text{Ph}\text{, v}}{i}_{\text{Ph},\text{v}}^2+R_{\text{Ph}\text{, w}}{i}_{\text{Ph}\text{, w}}^2$$

from equivalent resistances in the strand with the respective phase R _ph and the respective phase currents z _ph,uvw . Equation (2) is based on the assumption that the individual energy stores have an identical state of charge and therefore no equalizing currents flow when connected in parallel. Compensating currents put additional strain on the batteries and thus lead to further losses.

für alle nachfolgenden Spannungslevel mittels Gleichung (2) vorausberechnet werden und unter diesen dann die Schaltvektoren mit den minimalen ohmschen Verlusten ausgewählt werden. Oft finden sich dabei mehrere Schaltkombinationen, welche rechnerisch die gleichen bzw. annähernd gleichen minimalen Verluste mit sich bringen. Für die Idealstromaufteilung i_effienz werden dann bspw. die Mittelwerte der jeweiligen Modulströme anhand von Permutationen möglicher Schaltvektoren mit minimalen Verlusten berechnet.The ideal current distribution is determined, for example, in Equation (1) by considering the ohmic losses that occur when implementing possible switching vectors $\overrightarrow{s}$

can be precalculated for all subsequent voltage levels using equation (2) and the switching vectors with the minimum ohmic losses can then be selected from these. There are often several switching combinations that mathematically result in the same or almost the same minimal losses. For example, the mean values of the respective module currents are then calculated for the ideal current distribution i _efficiency using permutations of possible switching vectors with minimal losses.

The aim of determining the offset current is to actively align the states of charge of the energy stores of the modules with one another, also referred to as English balancing, which advantageously results in an even discharge, correspondingly an even load on the energy stores and thus ultimately their longer service life and energy utilization. This goal is pursued by relieving less charged energy storage devices and charging more heavily charged energy storage devices. The loading or unloading is mapped via the offset currents i _offset . A positive offset current favors switching states that increase the load on the corresponding energy store, while a negative offset current favors switching states that reduce the load. In the following method step of the method according to the invention, efficiency currents of the ideal current distribution i _efficiency are then combined with the offset currents i _offset in a target current calculation step, known as the “Target Current Calculation” (TCC) block.

However, it is also conceivable that, as an alternative or in addition to the goal of balancing states of charge, a further goal is a uniform thermal load, for example based on a temperature measured in the respective modules. When relieving a module, it is also conceivable to specify the relevant target current of 0 A instead of choosing a value that is lower than the target currents in other modules. If a ripple current (smaller in terms of amount) is desired, this is obtained by controlling an alternating component of the respective offset currents. In the case of small ripple currents, the module current from the previous modulation step can also be used as the target current. It is also conceivable to use a “state of energy” (SoE) or a “state of health” (SoH) as criteria for defining the offset current.

beschrieben wird. Dabei ist e_pm(t) ein Fehler zum Schaltzeitpunkt t, welcher eine jeweilige Abweichung des SoC vom arithmetischen Mittel aller Ladezustände $\overline{SoC}$

darstellt. $$e_{pm}=So{C}_{pm}-\overline{SoC},$$

wobei Index p und Index m ein jeweiliges Modul im Strang p an Position m bezeichnet. Das arithmetische Mittel aller Ladezustände $\overrightarrow{SoC}$

berechnet sich aus $$\overrightarrow{SoC}=\frac{1}{3n_{\text{ph}}}{\displaystyle \sum {}_{p=1}^3{{\displaystyle \sum {}_{m=1}^{n_{ph}}SoC}}_{pm}},$$

wobei n_ph eine Zahl von Modulen im jeweiligen Strang p darstellt. Um einen Wertebereich bei einer Implementierung klein zu halten, wird eine dynamische Verstärkung, vergleichbar mit einer Normierung eingesetzt. Es erfolgt eine Division durch eine mittlere absolute Abweichung vom arithmetischen Mittel aller Ladezustände $d_{\overline{SoC}}\left(SoC\right),$

$$d_{\overline{SoC}}\left(SoC\right)=\frac{1}{3n_{\text{ph}}}{\displaystyle \sum {}_{p=1}^3{\displaystyle \sum {}_{m=1}^{n_{ph}}\left|So{C}_{pm}-\overline{SoC}\right|.}}$$

Mathematically, the additional loading or unloading can be represented, for example, by a P controller, which is formally defined as $$\mathrm{and}\left(t\right)=K_p\cdot e_{pm}\left(t\right)$$

is described. In this case, e _pm (t) is an error at switching time t, which is a respective deviation of the SoC from the arithmetic mean of all states of charge $\overline{SOC}$

represents. $$e_{pm}=SO{C}_{pm}-\overline{SOC},$$

where index p and index m designate a respective module in string p at position m. The arithmetic mean of all states of charge $\overrightarrow{SOC}$

calculates itself $$\overrightarrow{SOC}=\frac{1}{3n_{\text{Ph}}}{\displaystyle \sum {}_{p=1}^3{{\displaystyle \sum {}_{m=1}^{n_{pH}}SOC}}_{pm}},$$

where n _ph represents a number of modules in the respective string p. In order to keep a range of values small in an implementation, a dynamic amplification, comparable to normalization, is used. It is divided by a mean absolute deviation from the arithmetic mean of all states of charge ${\mathrm{i.e}}_{\overline{SOC}}\left(SOC\right),$

$${\mathrm{i.e}}_{\overline{SOC}}\left(SOC\right)=\frac{1}{3n_{\text{Ph}}}{\displaystyle \sum {}_{p=1}^3{\displaystyle \sum {}_{m=1}^{n_{pH}}\left|SO{C}_{pm}-\overline{SOC}\right|.}}$$

To prevent division by zero, a factor α with a limit a _lim is introduced: $$a=\{\begin{array}{c}{\mathrm{i.e}}_{\overline{SOC}},\text{f}\ddot{\text{and}}\text{right}{\mathrm{i.e}}_{\overline{SOC}}\left(SOC\right)>a_{lim}\\ a_{lim},\text{f}\ddot{\text{and}}\text{right}{\mathrm{i.e}}_{\overline{SOC}}\left(SOC\right)\le a_{lim}\end{array}.$$

The factor α puts the deviation of the SoCs into perspective and uses the mean absolute deviation from the arithmetic mean for this purpose. At the same time, the limit value a _lim can be used to set when the states of charge are sufficiently adjusted and the influence of balancing is reduced.

In order to obtain a current or offset current from the normalized deviation obtained in this way, a multiplication by is carried out a pointer length of the setpoint current pointer in d/q coordinates, which results in a complete calculation of the offset current i _offset,pm : $$i_{\text{offset}\text{,}pm}=\frac{1}{a}\cdot \left(SO{C}_{pm}-\overline{SOC}\right)\frac{\sqrt{i_{\text{q},\text{should}}^2+i_{\text{i.e}\text{,should}}^2}}{G_{\text{bal}}}.$$

The factor G _bal in the denominator determines the strength or aggressiveness of the balancing. The smaller the factor, the more dominant balancing becomes compared to efficiency optimization. Through heuristic experiments, the best results were obtained for a system with three phases, p=3, and five modules each, n _ph =5, for a value of G _bal =4.

Since the offset current is calculated using the above equations, the definition of the offset current is also referred to as an offset current definition.

According to the invention, during the recuperation detection, a check is made as to whether regenerative braking can be performed at a current operating point and energy can also be recovered in the process. For this purpose, an electrical power P _Motor,el of the electrical machine is estimated using machine equations, from the measured speed n and the d/q setpoint currents i _d,sp , i _q,sp , taking into account the ohmic losses in the motor windings. If the electrical power P _{Motor,el is} negative, a recuperative operating point is assumed. This binary information (P _Motor,el negative yes/no) is then processed in the TCC block, so that, for example, in motor operation, more heavily discharged energy stores are adjusted to less heavily discharged energy stores by unloading them. The energy stores are charged in a targeted manner only when the electric machine is operated in a regenerative or recuperative manner. If there is no recuperation, the target currents are limited below to 0 A and active charging is avoided in order to advantageously avoid unnecessary ohmic losses when actively recharging the energy store.

The target currents i _target are determined from the inputs to the TCC block. According to the invention, ideal current distribution and offset current definition are summed up and then limited by means of variable saturation. In this context, variable saturation means that the applicable limit values are variable and are adjusted accordingly during operation. Such limits are, in particular, maximum charging and discharging currents dependent on the state of the energy store, ie temperature and state of charge (SoC).

In addition to the dynamic distinction, there are fixed limits i _min and i _max according to the current carrying capacity of the energy storage. The target currents i _target are limited to protect the energy stores, with both discharging and charging currents being limited. However, despite this limitation, module currents that actually flow can be larger in terms of amount, since the implementation of the target currents is based on the cost function. Typically, the limit of the charging currents is lower than the respective module allows according to the technical information, e.g. according to a data sheet. A variable limitation of the charging currents, depending on the current SoC, is also conceivable. A Dynamic Charge Acceptance, abbreviated DCA, of the energy storage could be considered. In relation to the storage capacity, the DCA describes the extent to which an energy storage device is able to absorb an electrical charge. This depends on the temperature, the current state of charge, but also a load history of the energy storage device.

welcher die Zielströme am besten realisieren kann, ausgewählt und anschließend ausgegeben. Dazu werden anhand eines modellprädiktiven Ansatzes, bspw. aus voranstehend zitierter MMSPC-Beschreibung entnommen, für eine Auswahl an möglichen Folgezuständen auf den aktuellen Schaltzustand die Modulströme i_mod,pm vorausberechnet und entsprechende Werte der Kostenfunktion Jals eine quadratische Abweichung von den Zielströmen i_target bestimmt: $$J={\displaystyle \sum {}_{p=1}^3}{{\displaystyle \sum {}_{m=1}^{n_{ph}}\left(i_{target,pm}-i_{mod,pm}\right)}}^2.$$

Using the cost function, that switching vector $\overrightarrow{s}$

which can best realize the target streams is selected and then output. For this purpose, using a model-predictive approach, e.g. taken from the MMSPC description cited above, the module currents i _mod,pm are precalculated for a selection of possible subsequent states on the current switching state and corresponding values of the cost function J are determined as a quadratic deviation from the target currents i _target : $$J={\displaystyle \sum {}_{p=1}^3}{{\displaystyle \sum {}_{m=1}^{n_{pH}}\left(i_{ta\mathrm{right}Get,pm}-i_{mO\mathrm{i.e},pm}\right)}}^2.$$

mit einem kleinsten Wert J in Gleichung (9) realisiert, worauf dieser Schaltvektor $\overrightarrow{s}$

an die jeweiligen Module übermittelt wird. Während des Betriebes führt das erfindungsgemäße Verfahren vorteilhaft zu einem Angleichen der Ladezustände bei gleichzeitig effizientem Betrieb des Umrichters. Wenn die Ladezustände komplett angeglichen sind, dann ist die Effizienzoptimierung dominant und Verluste werden minimiert.The respective target currents i _target are best achieved by the switching vector $\overrightarrow{s}$

with a smallest value J in equation (9), whereupon this switching vector $\overrightarrow{s}$

is transmitted to the respective modules. During operation, the method according to the invention advantageously leads to an adjustment of the charge states with simultaneous efficient operation of the converter. If the states of charge are completely equal, then the efficiency optimization is dominant and losses are minimized.

Alternatively, instead of forming error squares in equation (9), it is conceivable to use a delta-sigma method to integrate the deviation between the converted actual currents and the target currents and to take this into account for the next switching cycle. $$i_{DS,pm}=i_{\text{target},pm}-i_{\text{model}\text{,opt}\text{,}pm}.$$

According to the invention, module currents are precalculated in real time or for each switching cycle of the multilevel converter.

• ▪ den jeweiligen Zielstrom basierend auf einer Sollstromvorgabe zu berechnen, indem
  • • von der Regelung taktsynchron zu einem Schalttakt des modularen Multilevelumrichters die Sollstromvorgabe als Sollspannungslevel für die mindestens eine Phase und ein Phasenwinkel bereitgestellt werden und von der Steuerung verlustminimale Idealströme zu einer Idealstromaufteilung berechnet werden,
  • • zusätzlich zu der Sollstromvorgabe ein Ladezustand für jedes Modul herangezogen wird und daraus für einen Ladezustandsausgleich innerhalb der Mehrzahl an Modulen lastbasierte Offsetströme zu einer Offsetstromfestlegung berechnet werden,
  • • Idealstromaufteilung und Offsetstromfestlegung aufsummiert und mittels vorgegebener Sättigungswerte begrenzt werden, wodurch die jeweiligen Zielströme in einer Zielstromaufteilung gebildet werden,
• ▪ und mittels einer Kostenfunktion diejenigen Schaltzustände der Mehrzahl an Modulen, welche die Zielstromaufteilung mit einer geringsten Abweichung realisieren, zu bestimmen und damit den jeweiligen Schaltbefehl pro Schalttakt zu bilden.

Furthermore, a system for a target current modulation of a modular multilevel converter is claimed, in which the system comprises an electrical load, a machine control, a modular multilevel converter with a plurality of modules and with a controller, each module of the plurality of modules having at least two controllable switches and has at least one energy store, at least one phase being formed by arranging the modules in a row and a respective phase for the electrical load being formed by the at least one phase, the controller being configured to send a respective switching command for each switching cycle, which includes all switching states to be set of the modules and thereby adjusts a respective current in the respective module of the plurality of modules, wherein the switching command is specified to provide a target current to be realized and to transmit to the plurality of modules, and to

• ▪ calculate the respective target current based on a target current specification by
  • • the setpoint current specification as a setpoint voltage level for at least one phase and one phase angle is provided by the control synchronously with a switching cycle of the modular multilevel converter and the controller calculates ideal currents with minimal losses for an ideal current distribution,
  • • in addition to the target current specification, a state of charge is used for each module and load-based offset currents are calculated from this for state of charge equalization within the plurality of modules to determine an offset current,
  • • Ideal current distribution and offset current definition are summed up and limited using specified saturation values, whereby the respective target currents are formed in a target current distribution,
• ▪ and using a cost function to determine those switching states of the plurality of modules that implement the target current distribution with the smallest deviation and thus form the respective switching command per switching cycle.

In one embodiment of the system according to the invention, the system is configured to carry out a method according to the invention.

Further advantages and refinements of the invention result from the description and the attached drawing.

It goes without saying that the features mentioned above can be used not only in the combination specified in each case, but also in other combinations or on their own, without departing from the scope of the present invention.

• 1 zeigt ein Ablaufschema zur Zielstrommodulation in einer Ausführungsform des erfindungsgemäßen Verfahrens.
• 2 zeigt ein Berechnungsschema zur Zielstrommodulation in der Ausführungsform des erfindungsgemäßen Verfahrens.

The figures are described in a coherent and comprehensive manner, and the same reference symbols are assigned to the same components.

• 1 shows a flowchart for target current modulation in an embodiment of the method according to the invention.
• 2 shows a calculation scheme for target current modulation in the embodiment of the method according to the invention.

In 1 a flowchart 100 for target current modulation 130 is shown in an embodiment of the method according to the invention. A target current i _target is calculated in a target current modulation 130 for each MMSPC module in a line with phase U 141 , phase V 142 and phase W 143 . The target current modulation 130 is arranged at the signal flow level after a regulation with a PI controller 110 and a delta-sigma modulator 120 and before a bus for switching command distribution 140 . First, a q-component setpoint current i _q,setpoint 101 and a d-component setpoint current i _d,setpoint 102 are converted by the PI controller 110 to a q-component setpoint voltage u _q 103 and a d-component setpoint voltage u _d 104 and then supplied to the target current modulation 130 by means of the delta-sigma modulator 120 as a desired voltage level lvl _out,uvw (designated lvl _o" t,", lvl _out,v , lvl _out,w ). Other input variables for target current modulation 130 are also the q component target current i _q,target 101 and d component target current i _d,target 102, phase currents i _ph,uvw 106, (rotor) speed n 107, and all charge states SoCs 108 of the energy storage . The target current modulation 130 provides a switching command 109 as an output variable.

welcher die angestrebten Zielströme am besten realisieren kann, bestimmt und als Schaltbefehl 109 ausgegeben.In 2 a calculation scheme 200 for target current modulation in the embodiment of the method according to the invention is shown. A modulation of the target current i _target is made up of an ideal current distribution 210 , an offset current determination 220 and a recuperation detection 230 . The starting point is always maximum efficiency. At any point in time, with the ideal current distribution 210 with input variables target voltage level lvl _out,uvw 105 and phase currents i _ph,uvw 106, an ideal current distribution 210 i _efficiency that is optimal from the point of view of efficiency is determined across all modules and a calculation of the target current distribution 240, “Target Current Calculation” (TCC). With this ideal current distribution, the lowest ohmic losses occur in the converter and the energy storage devices. The calculation of the offset current definition 220 with the input variables charge states SoCs 108, q-component target current z _q,target 101 and the d-component target current i _{d,target 102 is used to} compensate for charging states. The offset current distribution 220 i _offset obtained from this is also used to calculate the Target current sharing 240 supplied. Depending on whether a respective offset current is positive or negative, a load on the individual battery cells can be increased or decreased. The recuperation detection 230 determines from the input variables speed n 107, q-component setpoint current i _q,setpoint 101 and d-component setpoint current i _d,setpoint 102 whether recuperative braking is taking place at the current operating point and forwards this information to the calculation of the target current distribution 240 , which leads there to the fact that the respective negative offset currents are set to the zero value without any recuperation. When calculating the target current distribution 240, the target current i _target or the target current distribution 245 is generated from i _efficiency and i _offset . Finally, using a cost function 250, a switching vector $\overrightarrow{s},$

which can best realize the desired target currents is determined and output as a switching command 109.

Reference List

100

Flow chart for target current modulation

101

q-component set current i _q,set

102

d-component set current i _d,set

103

q-component set voltage u _q

104

d-component target voltage u _d

105

Target voltage level lvl _{out, etc}

106

phase currents i _ph,uvw,

107

speed n

108

State of charge SoCs

109

switching command

110

PI controller

120

modulator

130

target current modulation

140

switching command distribution

141

Multilevel converter string for phase U

142

Multilevel converter string for phase V

143

Multilevel converter string for phase W

200

Scheme target current modulation

210

ideal current sharing

220

offset current setting

230

regeneration detection

240

target current sharing

245

target current

250

cost function

QUOTES INCLUDED IN DESCRIPTION

This list of documents cited by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent Literature Cited

• US20180109202 [0002]
• DE 102011075429 A1 [0005]
• DE 102016125720 A1 [0006]
• US20210155100A1 [0007]

Claims (11)

Method for a target current modulation of a modular multilevel converter, in which the modular multilevel converter has a plurality of modules and is controlled by a controller, in which each module of the plurality of modules has at least two controllable switches and at least one energy store, by lining up the modules at least one phase (141, 142, 143) is formed and a respective phase for an electrical load is formed by the at least one phase (141, 142, 143), with the controller issuing a respective switching command (109) for each switching cycle, which includes all switching states of the modules to be set, is provided and transmitted to the plurality of modules, with the respective switching command (109) setting a respective current in the respective module of the plurality of modules, with the switching command specifying a target current (245) to be implemented is, wherein the respective target current (245) by the controller based on a target current specification (101, 102) is calculated by • the target current specification (101, 102) is provided as a target voltage level (105) for the at least one phase and one phase angle by a controller synchronous with a switching clock of the modular multilevel converter and the controller calculates ideal currents with minimal losses for an ideal current distribution (210), • in addition to the target current specification, a state of charge is used for each module and load-based offset currents are calculated from this for state of charge equalization within the plurality of modules for an offset current definition (220), • Ideal current distribution (210) and offset current determination (220) are summed up and limited using predetermined saturation values, whereby the respective target currents (245) are formed in a target current distribution (240), and using a cost function (250) those switching states of the plurality of modules which implement the target current distribution (240) with the smallest deviation, be determined and thus the respective switching command (109) per switching cycle is formed. procedure after claim 1 , where the saturation values are constrained towards numerically larger values by technical upper limits and towards numerically smaller values are limited to the zero value. Method according to one of the preceding claims, wherein when calculating the respective target current (245), negative values are also permitted if a recuperative state (230) is present, whereby the modules are based on an estimate of an electrical power of the electrical machine based on a measured speed (107) and the setpoint current specification are charged in a targeted manner. Method according to one of the preceding claims, wherein the saturation values are adjusted during operation. Method according to one of the preceding claims, wherein a common-mode voltage level is additionally applied when calculating the respective target current (245). procedure after claim 5 , with all common-mode voltage levels that can be impressed being taken into account in the cost function (250) for all switching states. Method according to one of the preceding claims, wherein all possible target currents (245) are calculated for a switching clock which follows the current switching clock and are also taken into account in the cost function (250). Method according to one of the preceding claims, wherein in the case of equivalent switching states with equal minimum losses stored in the characteristics map, a mean value of the associated module currents is calculated or an arbitrary selection of one of these equivalent switching states is made. Method according to one of the preceding claims, wherein the target voltage level (105) is formed by PI control (110) and subsequent delta-sigma modulation (120) on the basis of the target current specification (101, 102). System for a target current modulation of a modular multilevel converter, in which the system comprises an electrical load, a machine control, a modular multilevel converter with a plurality of modules and with a controller, each module of the plurality of modules having at least two controllable switches and at least one energy store wherein at least one strand (141, 142, 143) is formed by arranging the modules in a row and a respective phase for the electrical load is formed by the at least one strand (141, 142, 143), the controller being configured to provide a respective switching command (109) for each switching cycle, which includes all the switching states of the modules to be set, and as a result a respective current is set in the respective module of the plurality of modules, a target current (245) to be implemented being specified for the respective switching command (109). and transmit to the plurality of modules, and to ▪ calculate the respective target current (245) based on a target current specification by • the target current specification (101, 102) as a target voltage level (105) for the at least one phase and a phase angle being provided by the control synchronously with a switching cycle of the modular multilevel converter and by of the controller, loss-minimum ideal currents are calculated for an ideal current distribution (210), • in addition to the target current specification, a state of charge for each module is used and from this load-based offset currents are calculated for a state of charge equalization within the plurality of modules for an offset current definition (220), • ideal current distribution (210 ) and offset current determination (220) are summed up and limited by means of predetermined saturation values, whereby the respective target currents (245) are formed in a target current distribution (240), • and by means of a cost function (250) those switching states of the plurality of modules that the target current distribution (240 ) with the smallest deviation, can be determined and thus form the respective switching command (109) per switching cycle. system after claim 10 , wherein the system is configured to perform a method according to any one of claims 2 until 9 to execute.

Images