Method for operating a pulse inverter when supplying an electric machine and pulse inverter

Abstract

Method for operating a pulse inverter (1) when supplying an electric machine, wherein switching angles (20) of pulses of a pulse duration modulation method implemented by means of a modulation control (2) of the pulse inverter (1) for driving power semiconductors (4) are determined by means of an optimization method taking into account at least one parameter (15) and / or an operating state (16) of the electrical machine, wherein the optimization procedure is carried out by means of the modulation control (2), wherein a state of the electrical machine is estimated by means of a machine model (7), wherein the switching angles (20) are determined by means of the optimization procedure when a difference between the estimated state and an actual state of the electrical machine exceeds a predetermined threshold.

Description

[0001] The invention relates to a method for operating a pulse inverter when supplying an electric machine and to a pulse inverter.

[0002] In most applications, electric machines are powered by a pulse inverter, with pulse width modulators used to control the torque and speed of the electric machines using fixed and predefined switching patterns. For reasons of ease of implementation and low cost, these are usually space-vector or sine-triangle pulse-width modulation (SVPWM, SPWM).

[0003] A variable, flexible variation of a pulse pattern, selected according to the specific electrical machine and spanning a certain level of the electrical machine's output, allows for the efficiency-optimized operation of a drive system consisting of a pulse inverter and an electrical machine. Such a pulse pattern, which differs significantly from SVPWM and SPWM, can be referred to as an optimized pulse pattern. Here, the on and off angles of the pulses within the pulse pattern are determined in advance for a drive system with specific characteristics using an optimization algorithm and configured during operation of the drive system via lookup tables.

[0004] From US patent 6,359,414 B1, a method for controlling a reluctance machine, in particular a switched reluctance machine, using a torque controller is known, in which motor windings are switched on at a turn-on angle and switched off at a commutation angle depending on the rotor angle position, wherein the angles are determined as control parameters as a function of the speed from stored control coefficients read during operation of the reluctance motor, wherein the commutation angle remains at least constant or increases for each speed, starting with a maximum value at a maximum torque as the torque values ​​decrease, and subsequently with a preset nominal value for the winding current, which is calculated as a control parameter for the windings of the reluctance motor and sent to a current controller.

[0005] From Zih-Cing You and Sheng-Ming Yang, A Restarting Strategy for Back-EMF-Based Sensorless Permanent Magnet Synchronous Machine Drive, Energies 2019, Volume 12, Issue 9, 1818; doi:10.3390 / en12091818, an observer for sensorless control of an electric machine is known.

[0006] From EP 2 863 528 A1, an inverter is disclosed comprising a first capacitor with a first DC voltage, at least two half-bridge modules, each with two semiconductor switches connected in series and each having a center terminal, and a clock generator for providing control signals to operate the semiconductor switches of the half-bridge modules in switching mode, wherein the half-bridge modules are connected to the first capacitor and thus connected in parallel. It is provided that a filter module for providing a second DC voltage is connected to the center terminals of the half-bridge modules, and the clock generator is configured to generate the control signals such that the first DC voltage at the first capacitor is converted into the second DC voltage.

[0007] From CN 202059359 U, a stator flux linkage observer of a salient-pole synchronous motor based on an FPGA (Field Programmable Gate Array) is known. The FPGA is used to control an analog-to-digital converter (ADC) to obtain a three-phase voltage and current harvest signal from a system; the three-phase voltage and current harvest signals are subjected to analog-to-digital conversion and then converted into voltage and current values ​​in polar coordinates by a 3 / 2 conversion module implemented by the FPGA; the voltage and current values ​​in polar coordinates are inputted to a neural network module implemented using the FPGA to obtain a flux linkage value.The flux link value is converted by a D / A (digital-to-analog) converter and then used as an output observation signal of the stator flux linkage of the salient-pole synchronous motor; and the FPGA is used to generate drive timing sequences for the A / D converter and the D / A converter. The stator flux linkage observer can effectively increase the observation accuracy of the flux linkage and improve the reliability of system operation.

[0008] DE 10 2020 213 401 A1 describes a method for operating an electric drive, wherein the electric drive is powered by a power converter, wherein a DC link voltage and a DC link current are detected by means of high-resolution sensors, wherein phase currents generated by the power converter to power the electric drive are detected by means of high-resolution phase current sensors, and wherein, starting from the detected DC link voltage, the detected DC link current, the detected phase currents and estimated and / or detected phase voltages, a power converter efficiency and / or a power converter loss is determined.wherein a modulation method of the power converter and / or at least one parameter of the modulation method are selected taking into account at least the specific power converter efficiency and / or the specific power converter loss and are used to supply the electric drive.

[0009] DE 10 2010 064 104 A1 describes a device for controlling an electric machine arranged in a vehicle, wherein the vehicle has driven wheels and the electric machine generates a torque acting on the driven wheels, and wherein the electric machine has at least three phases which can be connected to an energy storage unit via controllable switching elements of an inverter. The device includes a setpoint unit configured to provide a torque setpoint for the torque to be generated by the electric machine. Furthermore, the device includes a criterion unit configured to provide an optimization criterion. In addition, the device includes a voltage vector determination unit configured to determine a voltage vector associated with the torque setpoint.Furthermore, the device includes a duty cycle determination unit configured to determine duty cycle values ​​for the switching elements based on the voltage vector and the optimization criterion. The device also includes a control unit configured to generate control signals for the switching elements based on the duty cycle values, so that the electric machine generates a torque corresponding to the target torque value.

[0010] EP 3 654 524 A1 describes a control unit for operating an electric motor. The control unit is configured to determine values ​​of different phase currents of the electric motor for 2Z different scanning angles, where Z is an integer equal to or greater than one. One or more pairs of the 2Z scanning angles are spaced apart by a spacing angle of π / 6z, where z is an integer equal to or greater than one. Furthermore, the control unit is configured to determine an estimated value of a fundamental component of the phase currents based on the determined values ​​of the phase currents for the 2Z different scanning angles and using a Clarke and / or Park transform. The electric motor can then be operated depending on the estimated value of the fundamental component of the phase currents.

[0011] DE 10 2011 003 352 A1 describes a method, a system, and a device for increasing the voltage utilization in a five-phase vector-controlled machine drive system, which uses third harmonic current injection to increase the torque and power output of a five-phase machine. For this purpose, a fundamental current angle of a fundamental current vector is optimized for each specific torque-speed operating point of the five-phase machine.

[0012] CN 1 10 707 988 A describes a special control system for suppressing harmonics by pulse width modulation.

[0013] The invention is based on the objective of creating a method for operating a pulse inverter when supplying an electric machine and a pulse inverter in which, in particular, flexibility in the application of the pulse inverter is improved.

[0014] The problem is solved according to the invention by a method with the features of claim 1 and a pulse inverter with the features of claim 5. Advantageous embodiments of the invention are set forth in the dependent claims.

[0015] In particular, a method for operating a pulse inverter when supplying an electrical machine is provided, wherein switching angles of pulses of a pulse duration modulation method implemented by means of a modulation control of the pulse inverter for driving power semiconductors are determined by means of an optimization method taking into account at least one parameter and / or an operating state of the electrical machine, wherein the optimization method is carried out by means of the modulation control.

[0016] Furthermore, in particular a pulse inverter is created, comprising a modulation control, wherein the modulation control is configured to execute a pulse duration modulation method for driving power semiconductors and to determine switching angles of pulses of the pulse duration modulation method by means of an optimization method taking into account at least one parameter and / or an operating state of the electrical machine.

[0017] The method and the pulse inverter make it possible to determine efficiency-optimized switching angles, i.e., turn-on and turn-off angles, for the pulse-duration modulated switching of the pulse inverter's power semiconductors as needed. This significantly increases the flexibility of using the pulse inverter. For example, it may be possible to use the pulse inverter for various drive systems, especially for different electric machines. Before its first use, the pulse inverter is typically in an unconfigured or generally configured state.Depending on the specific application, the pulse inverter can then be configured to the electrical machine to be supplied by carrying out the procedure described in this disclosure and thereby determining switching angles using the optimization procedure, under which energy efficiency is optimized in particular.

[0018] One advantage of the method and the pulse inverter is that the switching angles do not need to be calculated in advance and provided in the form of a reference table, but can be determined directly in the modulation control as needed, taking into account at least one parameter and / or the operating state of the electrical machine.

[0019] Parts of the device, in particular the modulation control, can be designed individually or collectively as a combination of hardware and software, for example as program code that runs on a microcontroller or microprocessor. However, it is also possible for parts to be designed individually or collectively as an application-specific integrated circuit (ASIC).

[0020] The optimization process is carried out in particular by means of an additional processor unit of the modulation control set up for this purpose, so that the optimization process can be carried out in parallel with a pulse duration modulation control, i.e. a generation of control signals for the power semiconductors of the pulse inverter.

[0021] When executing the procedure, at least one parameter and / or the operating state of the electrical machine serve in particular as input parameters of the optimization procedure, especially in the form of boundary conditions during optimization.

[0022] Exemplary parameters and / or operating conditions may include the following: the temperature of the electric machine, the inductance (e.g., expressed as Lq, Ld) of the electric machine, the operating point of the electric machine, and / or faults of the electric machine, such as partial demagnetization and / or a short circuit in some cells of the electric machine.

[0023] This method is particularly suitable for all electrical machines, such as permanent magnet synchronous machines and reluctance machines. In the automotive sector, the interior permanent magnet (IPM) machine, with a superposition of reluctance and PM torque, predominates; here, too, the method can be used to advantage.

[0024] The method and the pulse inverter can be used, in particular, in a vehicle. The vehicle is specifically a motor vehicle. However, the vehicle can also be any other land, rail, water, air, or spacecraft. An aircraft, for example, is a drone or an air taxi.

[0025] In one embodiment, the optimization process includes minimizing harmonic overtones in the phase voltages generated by the pulse inverter. This allows for increased, and in particular optimized, energy efficiency. For minimizing the harmonic overtones, a selective harmonic elimination (SHE) method can be used, for example. By determining switching angles at which the harmonics are minimized, a portion of the total power is reduced by the harmonics, thus increasing energy efficiency during the operation of the pulse inverter and the electric machine. When implementing the SHE method, at least one parameter and / or the operating state of the electric machine serve as input parameters for the optimization process, particularly in the form of boundary conditions.

[0026] In one embodiment, the optimization method is executed using at least one field-programmable gate array (FPGA). In particular, the modulation control for this purpose comprises at least one FPGA. This allows the optimization method to be directly encoded and executed in hardware. The FPGA is programmed or configured to perform the optimization function. It can be provided, in particular, that the FPGA is at least partially reconfigurable, meaning that it can be partially reconfigured to a different function even during runtime.It may also be provided that the at least one field-programmable gate array provides functionality for both modulation control and optimization procedures.

[0027] The system envisages estimating the state of an electric machine using a machine model. Switching angles are then determined using an optimization procedure when the difference between the estimated state and the actual state of the electric machine exceeds a predefined threshold. This allows for the determination of optimized switching angles to improve energy efficiency in response to changes in the actual state of the electric machine. The estimation is performed using an estimator or a control system observer that uses the machine model to estimate the state of the electric machine based on acquired measurements (rotational speed, electrical rotor angle, phase currents, etc.). The estimated state is then compared with the actual state determined using acquired measurements.For example, voltages across the electric machine (e.g., the d / q voltages) can be estimated using the voltage differential equations of the electric machine and compared with voltages measured at the individual windings (or d / q voltages calculated from these). If a certain difference between the voltages exceeds a predefined threshold, the optimization procedure is executed again to determine optimized switching angles. In another example, the temperature of the electric machine is compared with a target range, and the optimization procedure is executed if the temperature deviates from the target range. The machine model can be provided, for example, by a control system observer. Machine learning methods can also be used to train and / or parameterize the machine model.Currents and voltages serve as input variables, and the observer estimates a magnetic flux and a torque based on these, or, in a sensorless control system, a rotor position angle. Such an observer can be implemented using known methods, for example, based on Zih-Cing You and Sheng-Ming Yang, A Restarting Strategy for Back-EMF-Based Sensorless Permanent Magnet Synchronous Machine Drive, Energies 2019, Vol. 12, Issue 9, 1818; doi:10.3390 / en12091818.

[0028] In one embodiment, the optimization process incorporates a machine learning method. This allows optimized switching angles to be learned. An adjustment for subsequent iterations when determining the optimal switching angles (also denoted as α1, α2, ..., αn) is achieved using an adjustment factor K_EM. This factor reflects the varying machine saturation behavior (= f(I,T), where I is the current and T is the temperature of the electrical machine) between two or three points determined by the observer. Specifically, this factor is determined using a lookup table, depending on the load level of the electrical machine.

[0029] Further features of the pulse inverter design are described in the various embodiments of the method. The advantages of the pulse inverter are the same in each case as in the embodiments of the method.

[0030] The invention is explained in more detail below with reference to preferred embodiments and the figures. These show: Fig. 1 a schematic representation of an embodiment of the pulse inverter; Fig. 2a a schematic representation of optimized switching angles as a result of the optimization procedure; Fig. 2b a schematic representation to illustrate the switching angles in the modulation method.

[0031] In Fig. Figure 1 shows a schematic representation of an embodiment of the pulse inverter 1. The method for operating a pulse inverter 1 when supplying an electric machine is explained in more detail below using the pulse inverter 1 as an example.

[0032] The pulse inverter 1 comprises a modulation controller 2, gate drivers 3, and power semiconductors 4, which in the example shown are configured as three half-bridges. The modulation controller 2 drives the gate drivers 3 using a pulse-width modulation method. The gate drivers 3 each drive the power semiconductors 4, so that a DC link voltage VDC is converted into an AC voltage, in particular into a rotating magnetic field, which supplies an electric machine (not shown).

[0033] In the example shown, the modulation control 2 forms part of a (not shown) controlled system, where the modulation control 2 is given d- and q-voltages VqCmd, VdCmd as manipulated variables to control the speed and torque of the electric machine using vector control. Furthermore, the current electrical rotor angle γ and the current speed ω of the electric machine are also supplied to the modulation control 2.

[0034] In a function block 10, the modulation control 2 determines the next n switching angles 20 of a pulse pattern. These determined switching angles 20 always alternate between switch-on and switch-off angles for pulses on each of the windings supplying the electric machine. In a function block 11, the modulation control 2 uses the determined switching angles 20 and the current rotational speed ω to calculate timer values, which are then passed to a pulse pattern generator 12. Based on these timer values, the pulse pattern generator 12 generates a pulse pattern to control the gate drivers 3. Additionally, a dead-time compensation can be performed in a function block 13, where switching edges are shifted in time to compensate for the dead time of the power semiconductors 4.

[0035] The modulation control 2 is configured to determine the switching angles 20 of pulses of the pulse duration modulation method by means of an optimization procedure taking into account at least one parameter 15 and / or an operating state 16 of the electric machine. The at least one parameter 15 and / or the operating state 16 represent, for example, a temperature of the electric machine and / or an operating point of the electric machine.

[0036] The optimization process is executed in a function block 14 encompassed by function block 10. Specifically, the optimization process is performed directly in the modulation control 2, eliminating the need to rely on externally provided switching angles 20. This increases the flexibility of the pulse inverter 1's application. In particular, the optimization process allows the pulse inverter 1 to be configured on-site for an electric machine and a specific application. In addition to increased flexibility in using the pulse inverter 1, this also improves energy efficiency during operation of both the pulse inverter 1 and the electric machine, as optimized switching angles 20 can be determined as needed at any given time.

[0037] Functional block 14 includes in particular a dedicated processor unit which can execute the optimization procedure in parallel with other functionality of the modulation control 2.

[0038] The optimization procedure is explained below using an example of an optimization method for minimizing harmonic overtones in the string voltages generated by the pulse inverter. In principle, other optimization methods can also be used additionally or alternatively.

[0039] Function block 14 receives the following input parameters: a list of modulation indices MI, a matrix A, a vector B, a list of initial values ​​x0, and a cost function F. The matrix A takes into account at least one parameter 15 and / or the operating state 16 of the electric machine.

[0040] The modulation index MI specifies, in particular, a ratio of a fundamental amplitude of a phase voltage of the electrical machine to half the DC link voltage.

[0041] The function block 14 provides a list as input data, in which each of the modulation indices MI contained in the list is assigned switching angles 20 under which the cost function F is minimized. From the list provided as the result of the optimization procedure, the respective optimized switching angles 20 can then be read or determined for a modulation level of the electric machine expressed via the modulation index MI.

[0042] In the example shown, the cost function F includes a weighted total harmonic distortion (WTHD), which describes a weighted ratio of the harmonic overtones to the fundamental frequency.

[0043] An example of MATLAB program code for performing the optimization procedure for selected harmonics is listed below:

[0044] Optimized switching angles 20 are calculated using the MATLAB code listed above. This minimizes the weighted total harmonic distortion. The cost function F corresponds to the function obj fun5THDi (x). Boundary conditions for optimization are specified for the MATLAB function fmincon via matrix A and vector B. At least one parameter 15 and / or the operating state 16 of the electric machine are taken into account via the factor K_EM, by which matrix A is multiplied. K_EM can be a factor or a vector. K_EM reflects, in particular, a different machine saturation behavior (= f(I,T), where I is the current and T is the temperature of the electric machine) between two or three points, which are determined, in particular, by an observer. This observer is determined, in particular, using a lookup table, depending on the load level of the electric machine.

[0045] An exemplary result provided by the optimization procedure is shown schematically in the Fig. Figure 2a shows the curve for one string (or phase). The x-axis represents the modulation index MI. The y-axis shows the optimized switching angles 20. The curves shown alternately correspond to a switch-on or switch-off angle.

[0046] In the Fig. Figure 2b shows a schematic representation to illustrate the switching angles. Fig. 2b shows a phase voltage U generated via the modulation method a of a phase through which the electric machine is powered, over a complete period. A situation for a specific modulation index MI is shown. The switch-on and switch-off angles correspond to the angles α. i, where the ideal sine wave intersects the switching edges of the pulse duration modulation. Due to quarter-wave symmetry, the switching angles for angles >90° can be derived from the switching angles from 0 to 90°.

[0047] Depending on the parameter 15 and / or operating state 16 of the electric machine, the optimization procedure yields different switching angles 20 after minimizing the cost function F. The switching angles 20 are then each optimized taking into account at least one parameter 15 and / or the operating state 16.

[0048] It may be provided that the modulation control 2 includes at least one field-programmable gate array 5 ( Fig.1) which executes the optimization procedure. In particular, it may be provided that the field-programmable gate array 5 is reconfigured depending on the optimization procedure to be used and / or parameter 15 and / or operating state 16. In particular, it may be provided that the field-programmable gate array 5 is configured for the electrical machine to be powered, at least on its first use, for example by parameterizing the factor K_EM accordingly.

[0049] The pulse inverter 1 is provided to have an observer 6, which is configured to estimate the state of the electrical machine using a machine model 7. The modulation control 2 is configured to determine the switching angles 20 using the optimization procedure when a difference between the estimated state and an actual state of the electrical machine exceeds a predetermined threshold. For example, the temperature of the electrical machine or its inductances Lq and Ld may change. If the change exceeds the predetermined threshold, optimized switching angles 20 are determined again. If the optimization procedure is to be executed again, a corresponding control signal is transmitted from the observer 6 to the function block 14 to start the optimization procedure.

[0050] The machine model 7 of the observer device 6 can be provided using a machine learning method.

[0051] It may be possible to configure the modulation control 2 to use a machine learning method when executing the optimization procedure. In particular, a trained neural network may be used.

[0052] It may be provided that the determined switching angles 20 are stored in a lookup table for easier retrieval. The lookup table is then updated, in particular, when updated switching angles 20 have been determined by re-executing the optimization procedure. Reference symbol list 1 pulse inverter 2 Modulation control 3 Gate drivers 4 Power semiconductors 5 field-programmable gate array 6 Observer equipment 7 Machine model 10 Function block 11 Function block 12 Pulse pattern generator 13 Function block 14 Function block 15 parameters 16 Operating state 20 switching angles VDC intermediate circuit voltage VqCmd specified q-voltage VdCmd specified d-voltage A Matrix B vector F cost function MI Modulation Index U a String voltage x0 Starting values

Claims

[1] Method for operating a pulse inverter (1) when supplying an electric machine, wherein switching angles (20) of pulses of a pulse duration modulation method implemented by means of a modulation control (2) of the pulse inverter (1) for driving power semiconductors (4) are determined by means of an optimization method taking into account at least one parameter (15) and / or an operating state (16) of the electrical machine, wherein the optimization procedure is carried out by means of the modulation control (2), wherein a state of the electrical machine is estimated by means of a machine model (7), wherein the switching angles (20) are determined by means of the optimization procedure when a difference between the estimated state and an actual state of the electrical machine exceeds a predetermined threshold. [2] Method according to claim 1, characterized by, that the optimization procedure includes minimizing harmonic overtones in string voltages generated by the pulse inverter (1). [3] Method according to claim 1 or 2, characterized by , that the optimization procedure is carried out using at least one field-programmable gate array (5). [4] Method according to any of the preceding claims, characterized by that the optimization process includes the use of a machine learning method. [5] Pulse inverter (1), comprising: a modulation control (2), wherein the modulation control (2) is configured to a to execute pulse duration modulation methods for controlling power semiconductors (4) and to determine switching angles (20) of pulses of the pulse duration modulation method by means of an optimization method taking into account at least one parameter (15) and / or an operating state (16) of the electrical machine, and an observer device (6) wherein the observer device (6) is configured to estimate a state of the electrical machine by means of a machine model (7), wherein the modulation control (2) is configured to determine the switching angles (20) by means of the optimization method when a difference between the estimated state and an actual state of the electrical machine exceeds a predetermined threshold value. [6] Pulse inverter (1) according to claim 5, characterized by, that the modulation control (2) is configured to minimize harmonic overtones in string voltages generated by the pulse inverter (1) as part of the optimization procedure. [7] Pulse inverter (1) according to claim 5 or 6, characterized by , that the modulation control (2) comprises at least one field-programmable gate array (5) that executes the optimization procedure. [8] Pulse inverter (1) according to any one of claims 5 to 7, characterized by , that the modulation control (2) is configured to use a machine learning method when executing the optimization procedure.

Images