CN-114114977-B - Method and device for controlling an electronic converter in a technical system by means of artificial intelligence

Abstract

The invention relates to a method for operating a technical device having an electronic converter controlled by at least one control signal, comprising the steps of providing a supplied control signal time sequence of one or more successive original control signals with which the electronic converter should be operated, predicting a predicted control signal time sequence of one or more future control signals from the first control signal time sequence, modifying the supplied control signal time sequence according to a trainable data-based control signal model according to the supplied control signal time sequence and the predicted control signal time sequence, in order to obtain a modified control signal time sequence of one or more successive control signals, wherein the control signal model is trained to determine the modified control signal time sequence according to the supplied control signal time sequence and the predicted control signal time sequence, and operating the electronic converter according to the waveform of the modified control signal time sequence.

Inventors

• M. R. Rudolph
• D. Blala
• S. V. arauyo
• M. Jeep Turner

Assignees

• 罗伯特·博世有限公司

Dates

Publication Date

20260508

Application Date

20210826

Priority Date

20200827

Claims (20)

1. 1. A method for operating a technical device (6) having an electronic converter (2) controlled by at least one control signal, the method having the steps of: -providing (S1) a control signal waveform (S) that should be used to steer the electronic converter (2); -predicting a predicted control signal waveform (S t+1 ...S t+T ) from the provided control signal waveform (S) as a predicted future waveform of the control signal; modifying the provided control signal waveform (S) according to a trainable data-based control signal model in accordance with the provided control signal waveform (S) and in accordance with the predicted control signal waveform (S t+1 ...S t+T ) so as to obtain a modified control signal waveform (S'), Wherein the control signal model is trained to determine a modified control signal waveform (S') from the provided control signal waveform (S) and the predicted control signal waveform (S t+1 ...S t+T ); -manipulating the electronic converter (2) according to a modified control signal waveform (S').
2. 2. The method according to claim 1, wherein the control signal model is designed to determine the modified control signal waveform (S') further as a function of one or more of the operating variables of the electronic converter (2) and/or of the technical device (6) to be actuated, one or more operating characteristics of the electronic converter (2) and/or of the technical device (6) to be actuated, one or more system characteristics of the electronic converter (2) and/or of the technical device (6) to be actuated, and one or more system variables of the technical device (6), wherein the operating variables influence the operation of the technical device (6).
3. 3. The method of claim 2, wherein the operating parameters include one or more of a thermal resistance of the overall structure, a capacitance of a backup capacitor coupled to the electronic component, and a variance of the parameters.
4. 4. A method according to any one of claims 1 to 3, wherein the control signal model is constructed as a trainable data-based model.
5. 5. A method according to any one of claims 1 to 3, wherein the control signal model is configured as an artificial neural network.
6. 6. The method of claim 5, wherein the artificial neural network is a multi-layer perceptron or a recurrent neural network.
7. 7. A method according to any one of claims 1 to 3, wherein the provided control signal waveform (S) and the predicted control signal waveform (S t+1 ...S t+T ) for manipulating the electronic converter (2) are parameterized or defined by control signal parameters and/or the modified control signal waveform (S') is parameterized or defined by corresponding modified control signal parameters.
8. 8. Method according to claim 7, wherein the parameterization of the provided control signal waveform and/or the modified control signal waveform (S') is performed by a time period and a value of an electrical parameter assigned to the time period.
9. 9. The method of claim 8, wherein the electrical parameter is voltage or current.
10. 10. Method according to claim 7, wherein the parameterization of the provided control signal waveform and/or the modified control signal waveform (S') is performed by one or more parameters of a periodic manipulation.
11. 11. The method of claim 10, wherein the one or more parameters include a periodic frequency, a frequency and/or a pulse width modulation, a modulation rate, a duty cycle, a pulse duration, and/or a pulse shape.
12. 12. A method according to any of claims 1 to 3, wherein the prediction of the predicted control signal waveform (S t+1 ...S t+T ) is performed by means of a data-based prediction model trained for determining the predicted control signal waveform (S t+1 ...S t+T ) from the provided control signal waveform (S).
13. 13. The method of claim 12, wherein the data-based prediction model comprises a recurrent neural network, a state space model, a Sequence2Sequence model, or a NARXGP model.
14. 14. A method according to any one of claims 1 to 3, wherein the modification of the provided control signal waveform (S) is performed in accordance with the trainable data-based control signal model in accordance with the provided control signal waveform (S) and in accordance with a predicted control signal waveform (S t+1 ...S t+T ) also in accordance with one or more operating parameters of the technical device (6).
15. 15. The method according to claim 14, wherein the one or more operating parameters specify a state of a technical device (6) to be controlled in accordance with the predicted control signal waveform.
16. 16. A method for training a control signal model, wherein the control signal model is structured for determining a modified control signal waveform (S') from a provided control signal waveform (S) and from a control signal waveform (S t+1 ...S t+T ) predicted based on the provided control signal waveform (S), the method having the steps of: -providing training data sets comprising respectively a provided control signal waveform (S) and a predicted control signal waveform (S t+1 ...S t+T ) that should be used to steer the electronic converter (2); -training the control signal model according to a loss function such that the provided control signal waveform (S) and the predicted control signal waveform (S t+1 ...S t+T ) are mapped to the modified control signal waveform (S '), wherein the loss function depends on one or more behavior metrics of a technical device (6) when the converter (2) is manipulated with the modified control signal waveform (S').
17. 17. The method of claim 16, wherein the control signal model is established for use in the method of any one of claims 1 to 15.
18. 18. The method according to claim 16, wherein the one or more behavior metrics (VM) each characterize a characteristic of the technical device (6) that depends on one or more behavior parameters (VG) that account for a behavior of an electronic circuit of the technical device (6) that depends on the modified control signal waveform (S').
19. 19. Method according to claim 18, wherein the respective behavior measure (VM) is determined as a function of a predefined cost for evaluating the behavior Variable (VG).
20. 20. Method according to claim 18, wherein the one or more behavior parameters (VG) are determined by means of measurement and/or circuit simulation.

Description

Method and device for controlling an electronic converter in a technical system by means of artificial intelligence Technical Field The invention relates to the control of an electronic converter in a technical system by means of a control signal. The invention relates in particular to measures for adapting control signals in order to achieve improved operating characteristics of a technical system. Background An electronic converter is an electronic circuit with one or more converter devices, which for example comprise transistors, MOSFETs, IGBTs or the like, in order to provide defined electrical parameters for operating a technical system, such as an electric motor. The electronic converter is manipulated by means of control signals which are applied to the base or gate connections of one or more active converter devices. In many applications, the control signals each follow a fixedly predefined waveform for different events and are often used to carry out a transition from the off-state to the on-state in a defined manner. The control signal is generated by a control unit or other driving unit. These signals often have a stepped or ramp waveform in order to achieve the desired switching or transfer characteristics of the converter device. Operation of such a converter, such as in an electric traction system of a motor vehicle, has an effect on the operating state of the traction system. As such, the operating characteristics may include the range and components of the vehicle, and the aging conditions of the traction battery, electronic converter, and traction motor, for example. Disclosure of Invention According to the invention, a method for operating a technical device having an electronic converter controlled by at least one control signal, and a corresponding device and a corresponding technical system are proposed. Other embodiments are described in the various embodiments. According to a first aspect, a method for operating a technical device having an electronic converter controlled by a control signal is provided, the method having the following steps: -providing a control signal waveform that should be used to operate the electronic converter; -predicting a predicted control signal waveform from the provided control signal waveform as a predicted future waveform of the control signal; Modifying the provided control signal waveform according to a trainable data-based control signal model in accordance with the provided control signal waveform and in accordance with the predicted control signal waveform, so as to obtain a modified control signal waveform, -Wherein the control signal model is trained to determine a modified control signal waveform from the provided control signal waveform and the predicted control signal waveform; -manipulating the electronic converter according to the modified control signal waveform. Electronic systems typically have converters that are part of or operate electronic circuits. Such a converter, in particular a converter device, an active electronic device such as a converter, is controlled by a sequence of control signal values of a control signal waveform, which is provided by a control unit. The sequence of control signal values of the control signal waveform is referred to herein as a control signal time sequence. For reasons of simpler processing of the signals, the control signal time sequence is assumed to be time-discrete. Furthermore, the converter may also be operated in the same or different ways by more than one control signal waveform. The control signal waveform or the control signal time sequence for operating the converter may have a state transition or a state waveform, in particular a potential change or a current change in the form of an edge or a ramp, which causes a corresponding reaction in the downstream electronic circuit. In the case of an electronic converter, the control signal time sequence may have a state transition or a state waveform, in particular a potential change or a current change in the form of an edge or a ramp, which follows a periodicity and which causes a corresponding reaction in the technical device associated therewith. The control signal waveform, in particular the form of the state transitions (voltage steps, voltage ramps, etc.), can decisively determine the behavior of the converter and the technical device operated thereby. For example, the power loss and the interference of the technical device, the service life of the system, and the load on the converter device or the electronic converter can be determined decisively by the state waveform, state transition or state waveform of the control signal waveform or of the control signal time sequence over time. According to the above method, a data-based control signal model is provided in order to optimize the shape of the control signal waveform (waveform over time) for the manipulation of the electronic converter. The control signal model is trained for generating