Control method, control system, computer program product, and computer readable medium

Abstract

The invention relates to a control method for controlling an electric drive by means of a control system having a motor vehicle, comprising: calculating at least one input parameter of the electric drive by means of a recurrent neural network of a first logic device of the control system, calculating at least one operating parameter of the electric drive on the basis of the at least one theoretical input parameter by means of a feedforward neural network of a second logic device of the motor vehicle of the control system, and controlling the electric drive by the control device of the motor vehicle on the basis of the calculated at least one operating parameter. The invention further relates to a control system comprising a first logic device and a motor vehicle having a second logic device and a control device, for carrying out the control method, as well as to a computer program product and to a computer readable medium.

Description

Technical Field

[0001] This invention relates to a control method for controlling an electric drive device via a control system for a motor vehicle. Furthermore, this invention relates to a control system for implementing this control method within a motor vehicle. Background Technology

[0002] Currently, motor vehicles are increasingly being driven by electric drive systems. The performance of an electric drive system is primarily determined by a favorable knowledge base regarding its performance capabilities and the utilization of its existing performance potential, as well as the advantageous control of the electric drive system based on the conditions under which these capabilities are utilized.

[0003] Known electric drive devices are mostly limited in their performance capabilities by means of basic structural design, design for temperature limits, and / or design for the minimum lifespan sought.

[0004] Obtaining the actual state of the electric drive unit in a motor vehicle is impossible in most cases because a large number of measuring devices are necessary and the corresponding measurement and / or computation costs of a motor vehicle are known to be impossible, or at least not cost-effective, in most cases.

[0005] Most known control methods utilize measurement data from the electric drive system during its operation and control the electric drive system using this measurement data. A simpler design approach for motor vehicles involves acquiring the performance capabilities of the electric drive system during the development and / or design phases and then statically controlling the electric drive system using these initially acquired performance capabilities.

[0006] A method and system for determining motor conditions (especially temperature) and for controlling motors are known from printed document CN 115 276 488 A.

[0007] For known motor vehicles and control methods for electric drive systems, the following is generally disadvantageous: the performance capabilities of the electric drive system are not optimally utilized and / or high measurement and / or computational costs in the motor vehicle are necessary to enable near real-time state acquisition of the electric drive system in the motor vehicle. Summary of the Invention

[0008] Therefore, the object of the present invention is to eliminate, or at least partially eliminate, the aforementioned disadvantages in the prior art. In particular, the object of the present invention is to provide a control method that makes it particularly easy to control an electric drive device through a control system of a motor vehicle. Furthermore, the object of the present invention is to provide a control system, a computer program product, and a computer-readable medium for controlling an electric drive device of a motor vehicle.

[0009] The aforementioned task is solved according to the present invention. In particular, the task is solved by a control method for controlling an electric drive device via a control system of a motor vehicle. Furthermore, the task is solved by a control system of a motor vehicle for implementing the control method, by a computer program product, and by a computer-readable medium. Further features and details of the invention are derived from the specification and drawings. Here, the features described in association with the control method according to the invention are equally applicable to the control system according to the invention, the computer program product according to the invention, and the computer-readable medium according to the invention, and vice versa, so that the disclosures regarding the respective aspects of the invention are always mutually referenced or can be mutually referenced.

[0010] According to a first aspect of the invention, this task is solved by a control method for controlling an electric drive device via a control system of a motor vehicle, the control method comprising: - At least one input parameter of the electric drive device is calculated by means of a recurrent neural network of the first logic device of the control system. - At least one operating characteristic parameter of the electric drive unit is calculated based on at least one theoretical input parameter by means of a feedforward neural network of the second logic device of the motor vehicle control system. - The electric drive unit is controlled by the motor vehicle's control unit based on at least one calculated operating characteristic parameter.

[0011] The method steps described previously and subsequently may preferably (unless otherwise explicitly stated) be performed individually, jointly, once, multiple times, in parallel time and / or sequentially in any order. Naming as, for example, "first method step" and "second method step" does not determine the temporal order and / or priority. The preferred order of the method steps is set such that the method steps are performed in the listed order.

[0012] The electric drive unit should preferably be understood as an externally excited drive motor or a permanently excited synchronous motor for driving motor vehicles. The electric drive unit is preferably designed as a synchronous machine and / or an asynchronous machine.

[0013] The computation of at least one input parameter is achieved using a recurrent neural network. Preferably, the computation of at least one input parameter is additionally or alternatively achieved using a convolutional layer. This convolutional layer is preferably integrated into the recurrent neural network. A recurrent neural network is preferably understood as a neural network that, in contrast to a feedforward neural network, is characterized by connections from one layer of neurons to neurons in the same or previous layers. To illustrate intuitively and illustratively, recurrent neural networks consider temporal relationships and / or time series.

[0014] For recurrent neural networks to compute at least one input parameter, driving parameters provided for training are preferably obtained through offline simulation of a thermal network, particularly a lumped-parameter thermal network, and / or through measurement.

[0015] Within the scope of this invention, recurrent neural networks are preferably used to train input parameters for feedforward neural networks. The computation of at least one input parameter is preferably performed in time prior to the mass operation of the electric drive unit and / or motor vehicle. The transmission of at least one input parameter to the motor vehicle, particularly to a second logic device, is preferably performed once or at least only intermittently and / or non-persistently.

[0016] Intuitively speaking, this feedforward neural network preferably considers one, or more specifically, only one time dataset in turn.

[0017] The calculation of at least one operating feature parameter is implemented via a second logic device of the vehicle using a feedforward neural network. Preferably, the calculation of at least one operating feature parameter is additionally or alternatively implemented using a convolutional layer. This convolutional layer is preferably integrated into the feedforward neural network. The calculation of at least one operating feature parameter is preferably implemented, particularly in real time, during the operation of the vehicle. The calculation of at least one operating feature parameter is based on at least one pre-calculated input parameter.

[0018] The control of the electric drive unit is then achieved by the vehicle's control unit based on at least one calculated operating characteristic parameter.

[0019] The control method according to the invention is therefore particularly advantageous because the significant computational cost of the first method steps is diverted to the first logic device and the recurrent neural network. Consequently, the motor vehicle can more efficiently calculate at least one operating characteristic parameter and control the electric drive unit. The motor vehicle thus requires less computational power and / or can advantageously improve the calculation of at least one operating characteristic parameter and / or the control of the electric drive unit with the same computational power.

[0020] The advantageous division of the control method according to the invention into a recurrent neural network for training the input parameters of the feedforward neural network and a feedforward neural network for calculating the basis of control for the electric drive device enables extremely efficient use of the resources of the control system and, in particular, avoidance of drastic fluctuations in the results for at least one operating characteristic parameter.

[0021] To illustrate intuitively and illustratively, through a recurrent neural network (RNN), various measurements of an electric drive device, such as telemetry for measuring rotor temperature and thermocouples for measuring static component temperature, are used offline to train the RNN. This RNN optimizes the physical parameters of an embedded thermal network in a manner that relationships such as losses, heat capacity, and thermal resistance are learned through training cycles (Epoche, sometimes also called rounds). Therefore, it is preferable for the RNN not to train explicit equation parameters, such as polynomial coefficients for frictional losses or inherent losses, because the RNN learns these relationships in neurons and / or network layers based on the input parameters of the RNN.

[0022] Therefore, recurrent neural networks are preferably used offline to compute and learn the input parameters for the feedforward neural network. In the second logic device of the vehicle, exemplarily only the input parameters of the feedforward neural network are applied and at least one operational characteristic parameter is calculated therefrom. The feedforward neural network is preferably implemented in the vehicle's software.

[0023] To illustrate intuitively and illustratively, the input parameters used for recurrent neural networks and the resulting calculated input parameters for feedforward neural networks vary during training and application, driving their specific selection. For example, a feedforward neural network for a passively cooled motor with oil delivery to a transmission requires different calculated input parameters than a feedforward neural network for an oil-cooled motor using an oil pump. The choice and number of input parameters for recurrent neural networks and the resulting calculated input parameters for feedforward neural networks are variably designed during application and training.

[0024] With the control method according to the invention, the number of input parameters for the second logic device used to calculate at least one operating characteristic parameter is preferably significantly reduced, for example, in the controller software of a motor vehicle. To calculate at least one operating characteristic parameter, physical parameters and equation coefficients are unnecessary or at least minimally necessary because the calculated input parameters are considered by a recurrent neural network used for the calculation. The physical mapping of the electric drive device is preferably achieved through the rational selection of input parameters, training data, and network topology for the recurrent neural network.

[0025] This control method is particularly advantageous because the electric drive unit is made particularly easy to control through the control system of the motor vehicle, wherein preferably the computational cost of the motor vehicle is reduced and / or the control of the electric drive unit is improved.

[0026] According to a preferred improvement of the invention, in the case of the control method, the first logic device of the control system can be designed separately from the motor vehicle and / or as a fixed first logic device. The first logic device should preferably be understood as a computer device, server device, test bench, and / or computing center. The first logic device enables the calculation of at least one input parameter. The input parameter of the first logic device used to calculate at least one input parameter should preferably be generally understood as an influence parameter on the thermal performance and / or performance of the electric drive unit, and is described in detail below. The calculation of the recurrent neural network through the first logic device preferably includes training and / or learning of the parameters of the electric drive unit. The control method designed in this way is particularly advantageous because the design of the first logic device being separate from the motor vehicle and / or as a fixed first logic device makes it particularly easy to enable the control of the electric drive unit through a control system with the motor vehicle, wherein, preferably, the computational cost of the motor vehicle is reduced and / or the control of the electric drive unit is improved.

[0027] According to a preferred improvement of the present invention, in the case of the control method, the control method may be further configured as follows: - At least one drive parameter of the electric drive unit of the motor vehicle is measured by the measuring device of the control system, and / or - Simulation of at least one drive parameter of the electric drive system of a motor vehicle, particularly through a first logic device. The calculation of at least one input parameter and / or at least one operating characteristic parameter (BK) of the electric drive device is based on at least one measured and / or simulated drive parameter.

[0028] The input parameters of the first logic device used to calculate at least one input parameter should preferably be understood as actual or theoretical measurement data of the electric drive unit (e.g., test bench and / or test driving of the vehicle). The recurrent neural network is trained or learns the performance and / or thermal characteristics of the electric drive unit using measured or simulated drive parameters. The measuring device should preferably be understood as a sensor device at the electric drive unit and / or a sensor system with a large number of sensors, for example, in the form of a test bench. To illustrate intuitively and illustratively, the recurrent neural network of the electric drive unit is used offline for training the neural network using telemetry measurement techniques and thermocouples. The input parameters calculated from the measured and / or simulated drive parameters are calculated as operating characteristic parameters by a feedforward neural network using a second logic device of the vehicle. The operating characteristic parameters are used to control the electric drive unit via a control device.

[0029] Preferably, the measurement of at least one driving parameter is performed in advance during the offline training time of the recurrent neural network. At least one driving parameter is additionally or alternatively provided by a separate offline simulation, particularly a hot network, such as a lumped parameter hot network.

[0030] Preferably, the measured and / or simulated drive parameters used to calculate at least one operating characteristic parameter via the second logic device are at least partially replaced and / or supplemented by the vehicle parameters described below. This preferably occurs before at least one operating characteristic parameter for the immediate control of the drive unit is calculated.

[0031] The control method designed in this way is particularly advantageous because the electric drive unit is made particularly easy to control by a control system of a motor vehicle through the measurement and / or simulation of at least one drive parameter, wherein preferably the computational cost of the motor vehicle is reduced and / or the control of the electric drive unit is improved.

[0032] According to a preferred improvement of the present invention, in the case of the control method, the measurement and / or simulation of at least one driving parameter includes the measurement and / or simulation of at least one of the following values: - Input current of the electric drive unit, - Rotational speed of the electric drive unit, - Coolant temperature of the electric drive unit - Ambient temperature of the electric drive unit, - Thermal boundary conditions of electric drive devices.

[0033] As previously described, the input parameters of the first logic device used for calculating at least one input parameter should preferably be understood as inflow parameters relating to the thermal and / or performance characteristics of the electric drive device. The enumerated drive parameters are preferably measured data and / or analog parameters to facilitate advantageous calculations for at least one input parameter of the feedforward neural network. Preferably, the sensing and logic devices used to measure and / or simulate these input parameters are necessary only for providing input parameters for recurrent neural networks and therefore preferably no longer located at the vehicle.

[0034] Alternatively or additionally, the control method according to the invention includes the measurement and / or simulation of at least one hot spot of the electric drive device. In particular, the control method according to the invention includes the measurement and / or simulation of at least one hot spot of the electric drive device on both winding head sides of a permanently excited synchronous machine and / or at least one hot spot in the rotor, particularly in the hottest magnet of the rotor. Exemplarily, the control method according to the invention includes the measurement and / or simulation of the hottest rotor winding in the case of an externally excited synchronous machine and / or the hottest short-circuit bar in an asynchronous machine.

[0035] The control method designed in this way is particularly advantageous because the electric drive unit is made particularly easy to control by a control system of a motor vehicle through the measurement and / or simulation of at least one of the listed drive parameters, wherein preferably the computational cost of the motor vehicle is reduced and / or the control of the electric drive unit is improved.

[0036] According to a preferred improvement of the present invention, in the case of the control method, the calculation of at least one operating characteristic parameter includes the calculation of at least one actual and / or theoretical operating characteristic parameter, particularly at least one of the following characteristic parameters: - Maximum temperature of the electric drive unit, - Average temperature of the electric drive unit - Maximum performance of the electric drive unit - Average performance of the electric drive unit.

[0037] The advantage of using a recurrent neural network to calculate input parameters for a feedforward neural network is that not only practical but also theoretical operating characteristic parameters can be calculated. The input parameters are used to calculate operating characteristic parameters via the feedforward neural network. Within the scope of this invention, a practical operating characteristic parameter should be understood, for example, as the temperature value of a hot spot in the electric drive unit, which is interpolated between two measured values. This practical operating characteristic parameter is preferably physically measurable and / or reproducible at the electric drive unit. Conversely, a theoretical operating characteristic parameter within the scope of this invention should be understood, for example, as the average temperature distribution of the components of the electric drive unit, wherein this average temperature is not necessarily measurable at any location on the component in reality. Therefore, the theoretical operating characteristic parameter is not necessarily the actual condition or situation of the electric drive unit, but is explicitly defined as a theoretical operating characteristic parameter that is optimized for advantageous further calculations and / or control of the electric drive unit via the feedforward neural network. The theoretical operating characteristic parameter is preferably a simplification of physical facts, sufficiently accurate, which significantly makes further calculations possible and / or simplifies the control of the electric drive unit. Preferably, the control method includes calculating the average temperature for the stator core and / or rotor of the electric drive device.

[0038] To illustrate intuitively and illustratively, a major advantage of recurrent neural networks (RNNs) is the independent abstraction of the heat transfer relationship between the maximum node and a hypothetical average node, provided a sufficient data base is available. Traditional node networks can theoretically also compute this hypothetical average node; however, parameterization is significantly more complex because the abstract relationship is mapped via pre-defined composite characteristic curve entries. It is precisely in the case of optimizing multiple target nodes (especially maximum and average nodes) that traditional optimizers quickly reach their limits with inappropriate search intervals or initial parameters. Neural networks are not bound to fixed composite characteristic curve entries and can selectively approximate functional relationships through TNN architectures.

[0039] The generation of average and / or maximum values ​​is preferably ensured by deriving the average temperature from individual sensor measurements of the relevant components. This additional parameter is a label preferably trained onto the neural network in addition to the maximum value. Simply put, it generates additional predictive targets from the measurements. The TNN model thus obtains additional temperature nodes; all corresponding heat transfer parameters are preferably automatically identified during training.

[0040] In other words, the maximum value should preferably reflect the component's hot spot to ensure thermal limit temperature. Here, the focus is on component protection and temperature is incorporated into the higher-level derating strategy. Conversely, the average value should preferably be an effective value, which improves the control of the electric drive. The temperature dependence of electromagnetic parameters can be adapted through appropriate compensation measures (e.g., flux tracking). Thus, the torque model and, correspondingly, the efficiency are improved by the control method according to the invention.

[0041] The division of temperature thus leads to the improvement of its input parameters as a higher-level sub-function of temperature.

[0042] The control method designed in this way is particularly advantageous because the electric drive unit is made particularly easy to control by a control system of a motor vehicle through the calculation of at least one actual and / or theoretical operating characteristic parameter, wherein preferably the computational cost of the motor vehicle is reduced and / or the control of the electric drive unit is improved.

[0043] According to a preferred improvement of the present invention, in the case of the control method, the calculation of at least one input parameter is implemented as a pre-calculation that has been completed in time, and / or the calculation of at least one operating characteristic parameter and the control of the electric drive device are implemented during the operation of the motor vehicle.

[0044] The control method according to the invention is particularly advantageous when at least one input parameter is pre-computed in terms of computation time and, for example, passed to the second logic device and the feedforward neural network as a completed training dataset. The entire computational cost of the recurrent neural network and the first logic device can thus be advantageously utilized. The logic device of the vehicle, especially the vehicle's logic device, is therefore advantageously unburdened by this computational cost and can further operate directly with the input parameters computed from the recurrent neural network.

[0045] The feedforward neural network is preferably implemented during the operation of the vehicle via a second logic device and calculates at least one operating characteristic parameter of the electric drive unit based on at least one theoretical input parameter. The calculations of the feedforward neural network and the second logic device are preferably performed online and / or in real-time during the operation of the vehicle. This control method is particularly advantageous because the design scheme of calculating at least one input parameter and / or at least one operating characteristic parameter makes it particularly easy to enable the control of the electric drive unit by a control system integrated with the vehicle, wherein, preferably, the computational cost of the vehicle is reduced and / or the control of the electric drive unit is improved.

[0046] According to a preferred improvement of the present invention, in the case of the control method, the second logic device of the control system is designed to be located within the motor vehicle and / or as the controller of the motor vehicle.

[0047] The design of the second logic device within the motor vehicle and / or as a controller for the motor vehicle is an advantageous use of existing resources and particularly preferably enables the previously described implementation of the calculations of the feedforward neural network and the second logic device to be preferably performed online and / or in real time during motor vehicle operation. The design of the second logic device within the motor vehicle and / or as a motor vehicle controller further enables advantageous functional coordination with the use of additional motor vehicle parameters, as described below, for the calculation of at least one operating characteristic parameter and / or the control of the electric drive unit. The control method designed in this way is particularly advantageous because the design of the second logic device makes it particularly easy to enable the control of the electric drive unit through a control system of the motor vehicle, wherein, preferably, the computational cost of the motor vehicle is reduced and / or the control of the electric drive unit is improved.

[0048] According to a preferred improvement of the invention, in the case of the control method, the calculation of at least one operating characteristic parameter can be configured such that it is recursively and / or iteratively implemented. To illustrate intuitively and illustratively, the calculated operating characteristic parameter preferably flows into the subsequent temporal calculation of another operating characteristic parameter. Here, the two temporally misaligned operating characteristic parameters can be the same or different operating characteristic parameters. This control method is particularly advantageous because the design of the calculation of at least one operating characteristic parameter makes it particularly easy to enable the control of the electric drive unit through a control system with a motor vehicle, wherein, preferably, the computational cost of the motor vehicle is reduced and / or the control of the electric drive unit is improved.

[0049] According to a preferred improvement of the present invention, in the case of the control method, the control method may be configured as follows: - At least one vehicle parameter is provided to a second logic device and / or control device via an interface device of the vehicle, wherein the calculation of at least one operating characteristic parameter and / or the electric drive device is additionally implemented based on the control of at least one provided vehicle parameter.

[0050] The second logic device is defined as a component of the motor vehicle according to the invention. The provision and use of at least one motor vehicle parameter for the calculation of at least one operating characteristic parameter and / or the control of the electric drive unit is an advantageous full utilization of existing resources and information. For example, the second logic device is designed as a controller, as previously implemented. This controller, especially the controller of the electric drive unit, already includes a large amount of data and / or signals for the state monitoring, control, and / or manipulation of the electric drive unit. The integration of the motor vehicle parameter into the control method according to the invention via an interface device is therefore an advantageous complement to the functional range of the control method and / or control system. The control method designed in this way is particularly advantageous because the interface device makes it particularly easy to enable the control of the electric drive unit through a control system of the motor vehicle, wherein, preferably, the computational cost of the motor vehicle is reduced and / or the control of the electric drive unit is improved.

[0051] According to a second aspect of the invention, this task is solved by a control system comprising a first logic device and a motor vehicle with a second logic device and a control device. This control system is designed to implement the control method according to the first aspect. In the described control system, all the advantages of the control method according to the first aspect of the invention are realized. Preferably, the first logic device is designed separately from the motor vehicle as previously described. Preferably, the second logic device is designed as an integral part of the motor vehicle as previously described.

[0052] According to a third aspect of the invention, this task is accomplished by a computer program product for controlling an electric drive device via a control system of a motor vehicle. This computer program product includes instructions that cause the control system according to the second aspect to implement the method steps according to the first aspect. In the case of the described computer program product, all the advantages described for the control method according to the first aspect of the invention and / or for the control system according to the second aspect of the invention are realized.

[0053] The computer program product is preferably implemented as computer-readable instruction code in any suitable programming language such as JAVA, C++, C#, and / or Python. The computer program product is preferably stored on a computer-readable storage medium such as a data disk, a removable disk drive, volatile or non-volatile memory, or built-in memory / processor. The instruction code preferably influences or manipulates the computing unit of a computer or other programmable device, such as a control system, so that instructions are executed. Furthermore, the computer program product is preferably available on a network such as the Internet. The computer program product is preferably implemented not only by means of software but also by means of one or more dedicated electronic circuits (i.e., hardware) or in any hybrid form (i.e., by means of software components and hardware components).

[0054] According to a fourth aspect of the invention, this task is accomplished by a computer-readable medium on which a computer program product according to a third aspect is stored. In the case of the described computer-readable medium, all the advantages described relative to the control method according to the first aspect of the invention, the control system according to the second aspect of the invention, and / or the computer program product according to the third aspect of the invention are realized. Attached Figure Description

[0055] The control method, control system, computer program product, and computer-readable medium according to the present invention are explained in more detail below with reference to the accompanying drawings. They are schematically illustrated as follows: Figure 1 The control system, including the first logic unit and the motor vehicle, is shown in a side view. Figure 2 The previous view shows another control system with a first logic device and a motor vehicle. Figure 3 The flowchart illustrates the design scheme of the control method according to the present invention, and Figure 4 A computer-readable medium containing a computer program is displayed in a functional view. Detailed Implementation

[0056] Components with the same function and working method in Figures 1 to 4 The same reference numerals are provided accordingly in the accompanying drawings.

[0057] exist Figure 1 The control system 10, including the first logic device 12 and the motor vehicle 100, is schematically shown in a side view. For improved clarity, in... Figure 1 The accompanying drawings do not specify the method steps. The motor vehicle 100 includes an electric drive unit 110, a second logic unit 102, and a control unit 104. The first logic unit 12 is designed for the calculation 202 of at least one input parameter EP of the electric drive unit 110 using a recurrent neural network (RNN). The second logic unit 102 is designed for the calculation 204 of at least one operating characteristic parameter BK of the electric drive unit 110 based on at least one theoretical input parameter EP using a feedforward neural network (FNN). The control unit 104 is designed for the control 206 of the electric drive unit 110 based on the calculated at least one operating characteristic parameter BK. The first logic unit 12 of the control system 10 is separate from the motor vehicle 100 and designed as a fixed first logic unit 12.

[0058] exist Figure 2 The functional view shows another control system 10 with a first logic device 12 and a motor vehicle 100. For improved clarity, in... Figure 2The accompanying drawings do not specify the method steps. The motor vehicle 100 includes an electric drive unit 110, a second logic unit 102, and a control unit 104. The first logic unit 12 is designed for the calculation 202 of at least one input parameter EP of the electric drive unit 110 using a recurrent neural network (RNN). The second logic unit 102 is designed for the calculation 204 of at least one operating characteristic parameter BK of the electric drive unit 110 based on at least one theoretical input parameter EP using a feedforward neural network (FNN). The control unit 104 is designed for the control 206 of the electric drive unit 110 based on the calculated at least one operating characteristic parameter BK. Furthermore, the control system 10 includes a measuring device 14 designed for the measurement 208 of at least one drive parameter AP of the electric drive unit 110 of the motor vehicle 100. Additionally, the control system 10 is designed for the simulation 210 of at least one drive parameter AP of the electric drive unit 110 of the motor vehicle 100, here performed offline by the control system 10 via a thermal network. The calculation 202 of at least one input parameter EP is implemented based on at least one measured and / or simulated drive parameter AP. The calculation 202 of at least one input parameter EP is implemented as a pre-calculation that has been completed in time, and the calculation 204 of at least one operating characteristic parameter BK and the control 206 of the electric drive unit 110 are implemented during the operation of the motor vehicle 100. The second logic device 102 of the control system 10 is designed as the controller of the motor vehicle 100 within the motor vehicle 100.

[0059] exist Figure 3 The control method 200 according to the present invention is schematically illustrated in a flowchart. For improved clarity, in... Figure 3The accompanying drawings only illustrate the method steps. In a first method step, control method 200 includes the calculation 202 of at least one input parameter EP of the electric drive unit 110 by means of a recurrent neural network (RNN) of a first logic device 12 of the control system 10. In another method step, control method 200 includes the calculation 204 of at least one operating characteristic parameter BK of the electric drive unit 110 based on at least one theoretical input parameter EP by means of a feedforward neural network (FNN) of a second logic device 102 of the control system 10. In yet another method step, control method 200 includes the control 206 of the electric drive unit 110 by means of a control device 104 of the motor vehicle 100 based on the calculated at least one operating characteristic parameter BK. In another method step, control method 200 includes at least one drive parameter AP of the electric drive unit 110 of the motor vehicle 100 being measured 208 by the measuring device 14 of the control system 10 and / or simulated 210 of at least one drive parameter AP of the electric drive unit 110 of the motor vehicle 100, wherein the calculation 202 of at least one input parameter EP and / or the calculation 204 of at least one operating characteristic parameter BK of the electric drive unit 110 are based on at least one measured and / or simulated drive parameter AP. The simulated drive parameter AP is provided through offline simulation of control method 200. In another method step, control method 200 includes at least one vehicle parameter KP being provided 212 through the interface device 106 of the motor vehicle 100 to the second logic device 102 and / or control device 104, wherein the calculation 204 of at least one operating characteristic parameter BK and / or the control 206 of the electric drive unit 110 are additionally based on at least one provided vehicle parameter KP.

[0060] exist Figure 4 A computer-readable medium 400 with computer program product 300 is schematically shown in a functional view. The computer program product 300 is stored on the computer-readable medium 400. The computer program product 300 is designed to control an electric drive unit 110 (not shown) via a control system 10 (not shown) with a motor vehicle 100 (not shown). The computer program product 300 includes instructions that cause the control system 10 (not shown) to implement a control method 200 (not shown).

[0061] List of reference numerals 10 Control System 12 Logic Devices 14 Measuring device 100 motor vehicles 102 Second Logic Device 104 Control Device 106 Interface Device 110 Drive Unit 200 Control Methods 202 Calculation 204 Calculation 206 Control 208 Measurement 210 Simulation 212 provides 300 computer program products 400 Computer-readable media AP driver parameters BK running characteristic parameters EP Input Parameters KP Vehicle Parameters FNN feedforward neural network RNN (Recurrent Neural Network)

Claims

1. A control method (200) for controlling an electric drive (110) by means of a control system (10) of a motor vehicle (100), the control method (200) comprising: - a calculation (202) of at least one input parameter (EP) of the electric drive (110) by means of a recurrent neural network (RNN) of a first logic device (12) of the control system (10), - a calculation (204) of at least one operating characteristic (BK) of the electric drive (110) based on at least one theoretical input parameter (EP) by means of a feedforward neural network (FNN) of a second logic device (102) of the motor vehicle (100) of the control system (10), - a control (206) of the electric drive (110) by means of a control device (104) of the motor vehicle (100) based on the calculated at least one operating characteristic (BK).

2. The control method (200) according to claim 1, characterized in that, The first logic device (12) of the control system (10) is designed separately from the motor vehicle (100) and / or as a stationary first logic device (12).

3. The control method (200) according to any one of the preceding claims, characterized in that, The control method (200) furthermore comprises: - a measurement (208) of at least one drive parameter (AP) of the electric drive (110) of the motor vehicle (100) by means of a measurement device (14) of the control system (10), and / or - a simulation (210) of at least one drive parameter (AP) of the electric drive (110) of the motor vehicle (100), in particular by means of the first logic device (12), wherein the calculation (202) of the at least one input parameter (EP) and / or the calculation (204) of the at least one operating characteristic (BK) of the electric drive (110) is effected based on at least one measured and / or simulated drive parameter (AP).

4. The control method (200) according to claim 3, characterized in that, The measurement (208) and / or simulation (210) of the at least one drive parameter (AP) comprises a measurement (208) and / or simulation (210) of at least one of the following values: - an input current of the electric drive (110), - a rotational speed of the electric drive (110), - a coolant temperature of the electric drive (110), - an ambient temperature (110) of the electric drive (110), - a thermal boundary condition of the electric drive (110).

5. The control method (200) according to any one of the preceding claims, characterized in that, The calculation (204) of the at least one operating characteristic (BK) comprises a calculation of at least one actual and / or theoretical operating characteristic (BK), in particular at least one of the following operating characteristics (BK): - a temperature maximum of the electric drive (110), - a temperature average of the electric drive (110), - a performance maximum of the electric drive (110), - a performance average of the electric drive (110).

6. The control method (200) according to any one of the preceding claims, characterized in that, The calculation (202) of the at least one input parameter (EP) is realized as a precalculation which has been completed in time, and / or the calculation (204) of the at least one operating characteristic variable (BK) and the control (206) of the electric drive (110) are realized during operation of the motor vehicle (100).

7. The control method (200) according to any one of the preceding claims, characterized in that, The second logic device (102) of the control system (10) is designed in the motor vehicle (100) and / or as a controller of the motor vehicle (100).

8. The control method (200) according to any one of the preceding claims, characterized in that, The calculation (204) of the at least one operating characteristic variable (BK) is realized recursively and / or iteratively.

9. The control method (200) according to any one of the preceding claims, characterized in that, The control method (200) furthermore comprises: - the provision (212) of at least one motor vehicle parameter (KP) to the second logic device (102) and / or the control device (104) by means of an interface device (106) of the motor vehicle (100), wherein the calculation (204) of the at least one operating characteristic variable (BK) and / or the control (206) of the electric drive (110) are additionally realized on the basis of at least one provided motor vehicle parameter (KP).

10. A control system (10) comprising a first logic device (12) and a motor vehicle (100) with a second logic device (102) and a control device (104), characterized in that The control system (10) is designed to carry out the control method (200) according to any one of the preceding claims.

11. A computer program product (300) for controlling an electric drive (110) by a control system (10) with a motor vehicle (100), characterized in that The computer program product (300) comprises instructions which cause the control system (10) according to claim 10 to carry out the method steps according to any one of claims 1 to 9.

12. A computer-readable medium (400) on which the computer program product (300) according to claim 11 is stored.

Images