Method for estimating an excitation current in a separately excited electrical machine, device, reference device and system

Abstract

The present invention provides a method for estimating an excitation current (I_R) in a separately excited electrical machine (10), comprising the steps of: providing (S1) an output voltage by a voltage source (14); converting (S2), by an inverter (22), the output voltage (U_A) into a stator voltage (U, V, W), which is applied to a stator (11) of the electrical machine (10) for controlling a torque and / or a rotational speed of a rotor (12) of the electrical machine (10); impressing (S3) an excitation current (I_R) into the rotor (12) by means of contactless energy transfer from a primary current generated from the output voltage; detecting (S4), with at least one sensor (13), measurement data at the stator (11) to provide input data, wherein the measurement data contain at least a terminal voltage, the rotational speed of the rotor (12), an active current, and a reactive current;and calculating (S5), with a processor (231) of a control device (23), the excitation current (I_R) using artificial intelligence based on the input data and training data. The present invention further provides a corresponding device (20) for a vehicle (100), a reference device (30), and a system (200) comprising the inventive device (20) and the reference device (30).

Description

[0001] The present invention provides a method for estimating an excitation current in a separately excited electrical machine, a device for a vehicle, a reference device for determining training data for a method for estimating an excitation current in a separately excited electrical machine and a system for estimating an excitation current in a separately excited electrical machine in a vehicle. State of the art

[0002] In the past, the output current of a separately excited electric motor (ESM) or separately excited synchronous motor (SJM) was transmitted to the rotor via slip rings. In this case, the current can be measured using a sensor on the stator. In newer separately excited electric motors, the current is transmitted to the rotor using contactless energy transfer, for example, with an inductive coupling. In this case, the current on the rotor can no longer be measured directly. A current sensor on the rotor and data transmission are difficult to implement due to the high temperatures and the associated costs.While it is possible to estimate the excitation current in the rotor using a physical observer based on the available measured variables (stator currents, stator voltage, speed, position), nonlinear changes in the machine parameters, particularly in the inductances, which are typically dependent on the current flow, and in the rotor resistance, which is typically dependent on the temperature, lead to complex physical observer structures. This can cause inaccuracies in the estimation of the output current, which usually limit the operating range of the electric motor. This then leads to the electric machine no longer being fully utilized in its intended speed or torque range.

[0003] DE 11 2020 007 296 T5 describes an estimation device for a motor temperature and a torque. Disclosure of the invention

[0004] The present invention provides a method for estimating an excitation current in a separately excited electrical machine according to claim 1, a device for a vehicle according to claim 12, a reference device for determining training data for a method for estimating an excitation current in a separately excited electrical machine in a vehicle according to claim 13, and a system for estimating an excitation current in a separately excited electrical machine in a vehicle according to claim 14.

[0005] Preferred further training is the subject of the subclaims. Advantages of the invention

[0006] According to the invention, the method for estimating an excitation current in a separately excited electrical machine comprises the steps of: providing an output voltage by a voltage source, converting, by an inverter, the output voltage into a stator voltage which is applied to a stator of the electrical machine for controlling a torque and / or a speed of a rotor of the electrical machine, impressing an excitation current into the rotor by means of contactless energy transfer from a primary current generated from the output voltage, detecting, with at least one sensor, measurement data on the stator to provide input data, wherein the measurement data contain at least a terminal voltage, the speed of the rotor, an active current and a reactive current, and calculating, with a processor of a control device, the excitation current by means of artificial intelligence based on the input data and training data.

[0007] According to the invention, the device for a vehicle comprises a voltage source for providing an output voltage; a separately excited electrical machine having a stator and a rotor; a coupler configured to impress a primary current generated by the output voltage into an excitation current in the rotor by contactless energy transfer; an inverter for controlling a torque and / or a rotational speed of a rotor of the electrical machine by converting the output voltage into a stator voltage; at least one sensor for acquiring measurement data at the stator of the electrical machine, the measurement data including at least a terminal voltage, a rotational speed of the rotor, an active current, and a reactive current; and a control device connected to the electrical machine and the at least one sensor, and having a processor and a non-volatile memory.wherein the memory contains computer-readable instructions enabling the processor to calculate the excitation current using artificial intelligence based on the measurement data as input data and training data.

[0008] According to the invention, the reference device for determining training data for a method for estimating an excitation current in a separately excited electrical machine, in particular for a method according to the invention, comprises a device according to the invention, a telemetry sensor arranged in the rotor, which is arranged and designed to detect the impressed excitation current; a data processing device connected to the telemetry sensor, which is designed to operate the electrical machine over an operating range in order to thereby determine reference excitation current data from the detected excitation current and reference measurement data from the detected measurement data, and to provide training data for an artificial intelligence based on the determined reference measurement data and the reference excitation current data.

[0009] The idea underlying the present invention is therefore to provide a data-based model for determining the unknown and only measurable excitation current of a separately excited electrical machine (ESM) or a separately excited electric motor. This model thus helps to reduce the number of components in an electric motor.

[0010] With the data-based model, the invention thus creates a virtual sensor that is implemented using a data-based model. The virtual sensor is thus understood to be the computer-readable instructions on a memory which, together with measurement data supplied by at least one sensor, provide the excitation current in the rotor of the ESM. The measurement data includes at least a terminal voltage, a rotor speed, an active current, and a reactive current at the stator. These variables are to be understood in the commonly used sense of an electric motor and can be determined either directly or via the stator voltage, which is typically a three-phase alternating voltage and generates a three-phase current to operate the ESM. The output voltage for generating a primary current can conveniently be provided by a battery arranged in the vehicle.The contactless energy transfer of the primary current to the rotor can be carried out on a shaft of the rotor of the electric motor or electric machine.

[0011] The virtual sensor uses a feedforward network that does not use the predicted output signal as the input signal for the next time step, but rather predicts the output signal from the signals of the current time step. The measured data forms the input signal, and the calculated or estimated excitation current forms the output signal.

[0012] The virtual sensor is executed directly on the control unit using an embedded algorithm, or embedded code. The inverter is preferably used to execute the algorithm or artificial intelligence. However, it is also conceivable to execute the algorithm or artificial intelligence on a suitable separate control unit. This could be, for example, on a central control unit of the vehicle or in a cloud with a real-time connection to the vehicle.

[0013] The training, which generates the training data, is carried out in such a way that labeled data can be recorded on a special sensor system for current detection on the rotor via telemetry data transmission. The excitation current on the rotor is thus measured directly.

[0014] To adequately estimate the excitation current based on the training data, at least the above-mentioned signals or measurement data are required as input data for the artificial intelligence model: a terminal voltage, a machine speed, an active current, and a reactive current, which are to be understood in the sense of conventional electric motors. These variables can be detected by at least one sensor, preferably by a sensor dedicated to each signal. The sensor(s) are arranged on or adjacent to the stator, thus eliminating the need for a sensor on the rotor.

[0015] By connecting, for example via a cloud, to the reference system or the reference device, the data-based model and the training data in the vehicle can be updated and adapted so that current training data is always available for a precise prediction of the excitation current in the rotor.

[0016] According to a preferred embodiment of the method, the primary current is an AC primary current. The contactless energy transfer is carried out by means of inductive coupling in a coupler. An AC excitation voltage is induced in a rotor section of the coupler arranged in the rotor by the AC primary current in a primary section of the coupler. An AC excitation current induced from the AC primary current is then rectified into the excitation current by a rectifier in the rotor. This represents a particularly efficient implementation of contactless energy transfer.

[0017] According to a preferred embodiment of the method, the at least one sensor comprises a plurality of sensors. Of the plurality of sensors, at least one current sensor is arranged and configured to measure the active current and the reactive current in the inverter.

[0018] According to a preferred embodiment of the method, the artificial intelligence contains at least one Gaussian process algorithm, a neural network, or a multi-layer perceptron algorithm. Thus, the virtual sensor according to this invention can be implemented using a neural network (NN), a Gaussian process (GP), or multi-layer perceptron (MLP). In the case of a Gaussian process algorithm, uncertainty information contained in the algorithm can be used as a criterion for validating the algorithm used (AI model). These algorithms are therefore particularly suitable for this application, allowing an estimate of the excitation current to be achieved with high accuracy and sufficiently low computing time.

[0019] According to a preferred embodiment, the method comprises the additional step of calculating a physical quantity from the measured data. The input data comprises the calculated physical quantities. In the simplest case, the calculated physical quantities can be the calculation of a voltage from a measured current and a known resistance (U=R·I). Thus, physical information is introduced into the artificial intelligence, which allows the artificial intelligence to operate more efficiently, thus saving computing time and increasing the accuracy of the estimation.

[0020] According to a preferred embodiment of the method, the control device is arranged within the inverter and configured to convert the stator voltage from the output voltage. Since the control device receives the measurement data anyway, this implementation is particularly compact and efficient.

[0021] According to a preferred embodiment of the method, the output current generated by the output voltage is a direct current, which is converted into the primary current by a second inverter. In this case, the output voltage is a direct voltage, which can be provided, for example, by a battery as a voltage source. The direct current generated by the battery is converted by the second inverter into an alternating voltage, which is used for contactless energy transfer, in particular in the form of an inductive coupling, to inject the excitation current into the rotor of the electric machine.

[0022] According to a preferred embodiment of the method, the measurement data further includes a temperature, the primary current, a direct voltage, in particular the output voltage, and / or a direct current generated, in particular, by the output voltage. Additional parameters as input data increase the accuracy of the estimation. The temperature can, for example, relate to the temperature of a part of the rotor, the stator, or components used for inductive coupling.

[0023] According to a preferred embodiment of the method, the training data is determined from reference measurement data acquired by a telemetry sensor. The telemetry sensor is arranged and configured to detect the excitation current of a rotor of an electrical machine of a reference device, induced by contactless energy transfer. Alternatively or additionally, the training data is determined by a computer simulation based on a physical model of the separately excited electrical machine. This provides training data suitable for the application, which enables an estimation of the excitation current with sufficient accuracy for the application.

[0024] According to a preferred embodiment, the method comprises the further step of updating the training data based on newly determined reference measurement data, which was determined using the electrical machine of the reference device. This provides a correction of the current data. Only a relatively small number of machines, i.e., separately excited electrical machines or electric motors installed in vehicles or systems, are equipped with corresponding (telemetry) sensors that provide such reference measurement data. Thus, the virtual sensor is updated over its lifetime, thus continuously improving the estimate of the excitation current.

[0025] According to a preferred embodiment, the method comprises the further step of controlling, with the inverter or the control device, the electric machine based on the calculated excitation current to set a predetermined speed and / or a predetermined torque. Based on the estimated excitation current, a speed range or a torque range of the electric motor can thus be utilized particularly effectively, i.e., almost fully.

[0026] According to a preferred embodiment of the system, the vehicle is included in a vehicle fleet. Each vehicle in the vehicle fleet contains a device according to the invention integrated into the respective vehicle. The data processing device is designed to transmit the update to each vehicle in the vehicle fleet. In this way, the artificial intelligence can be continuously improved in each vehicle in the vehicle fleet.

[0027] According to a preferred embodiment of the system, the electric machine of the reference device is arranged in a vehicle. The data processing device is configured to provide training data based on measurement data and excitation current data determined during vehicle operation. In this way, training data can be obtained during vehicle operation under realistic road conditions, which contribute to a more accurate estimation of the excitation current.

[0028] Further features and advantages of embodiments of the invention will become apparent from the following description with reference to the accompanying drawings. Short description of the drawings

[0029] The present invention is explained in more detail below with reference to the exemplary embodiments shown in the schematic figures of the drawing.

[0030] They show: Fig. 1 a schematic representation of a sequence of method steps of a method for estimating an excitation current in a separately excited electrical machine according to an embodiment of the present invention; Fig. 2 is a schematic representation of a device for a vehicle according to an embodiment of the invention; Fig. 3 a schematic representation of an algorithm of an artificial intelligence applicable in the method according to the invention and the control device according to the invention; Fig. 4 is a schematic representation of a method for determining training data for a method for estimating an excitation current in a separately excited electrical machine according to an embodiment of the present invention; and Fig. 5 is a schematic representation of a system for estimating an excitation current in a separately excited electrical machine in a vehicle according to an embodiment of the present invention.

[0031] In the figures, the same reference symbols denote the same or functionally identical elements.

[0032] Fig. 1 shows a schematic representation of a sequence of method steps of a method for estimating an excitation current I_R in a separately excited electrical machine 10 according to an embodiment of the present invention.

[0033] In the method, first, an output voltage U_A is provided S1 by a voltage source 14. Furthermore, an inverter 22 converts the output voltage U_A into a stator voltage U, V, W S2. The stator voltage is applied to a stator 11 of the electric machine 10 to control a torque and / or a rotational speed φ of a rotor 12 of the electric machine 10. Furthermore, an excitation current I_R is impressed S3 into the rotor 12 by means of contactless energy transfer from a primary current generated from the output voltage U_A.

[0034] Furthermore, at least one sensor 13 acquires measurement data on the stator 11 to provide input data S4. The measurement data includes at least a terminal voltage, the rotational speed φ of the rotor 12, an active current, and a reactive current. These variables are to be understood in the sense of a conventional electric motor 10. Furthermore, the measurement data can include a direct voltage, U_dc, such as the output voltage U_A, and / or a direct current, for example the direct current I_dc generated by a direct current voltage source 14. The measurement data can also include a temperature T_S, which is acquired, for example, from a part of the stator 11 or rotor 12.

[0035] The excitation current I_R is now calculated S5 using artificial intelligence based on the input data and training data using the processor 231 of the control unit 23. The functionality of the artificial intelligence is described below with reference to Fig. 4. Finally, a predetermined rotational speed φ and / or a predetermined torque can be set based on the calculated excitation current using the inverter 22 or the control device 23 arranged in the inverter 22.

[0036] Fig. 2 shows a schematic representation of a device 20 for a vehicle 100 according to an embodiment of the invention.

[0037] The device 20 comprises a voltage source 14 for providing an output voltage U_A. In preferred embodiments, the output current I_dc generated by the output voltage U_A is a direct current. The output current I_dc is then converted by a second inverter 25 into a primary current I_P, which in these embodiments is an AC primary current, i.e., an alternating current.

[0038] The device 20 further comprises a separately excited electric machine 10, which has a stator 11 and a rotor 12. In preferred embodiments, the electric machine 10 is designed as an electric motor 10 for a vehicle 100. The rotor 11 has a rotor body 12a, which is arranged within the stator 11 and rotatably mounted.

[0039] The device 20 also comprises a coupler 24 which is designed to impress a primary current I_P generated by the output voltage U_A into an excitation current I_R in the rotor 12 by contactless energy transfer.

[0040] The primary current is preferably an AC primary current, and the contactless energy transfer is preferably carried out by means of inductive coupling in the coupler 24. An AC excitation voltage is induced in a rotor section 12b of the coupler arranged in the rotor 12 by the AC primary current I_P in a primary section 24a of the coupler 24. An AC excitation current induced from the AC primary current I_P is rectified into the (DC) excitation current I_R by a rectifier 12c, which is preferably a diode, in the rotor 12. The inductive coupling, or mutual induction, is realized by two closely arranged circuits, or conductor loops. These can, for example, contain coils in an arrangement similar to a transformer. The rotor section 12b is arranged within the rotor body 12a.

[0041] The device 20 also comprises an inverter 22 for controlling a torque and / or a rotational speed φ of the rotor 12 of the electric machine 10 by converting the output voltage U_A into a stator voltage U, V, W. The stator voltage U, V, W is preferably a three-phase alternating current with the phases U, V and W.

[0042] The device 20 further comprises at least one sensor 13 for acquiring measurement data on the stator 11 of the electric machine 10. The measurement data includes at least a terminal voltage, a rotational speed φ of the rotor 12, an active current, and a reactive current. Furthermore, the device 20 can comprise a plurality of sensors 13. Of the plurality of sensors 13, at least one current sensor for measuring the active current and the reactive current is arranged and configured in the inverter 22. This also includes the possibility of determining the active current and the reactive current from the values ​​measured by the sensor 13, such as the phases U, V, and W of the stator voltage.

[0043] Furthermore, the device 20 comprises a control device 23, which is connected to the electric machine 10 and the at least one sensor 13. The control device 23 has a processor 231 and a non-volatile memory 232. The memory 232 contains computer-readable instructions with which the processor 231 is able to Fig. 1 described calculation of the excitation current I_R in the electrical machine 10.

[0044] In this exemplary embodiment, the control device 23 is arranged within the inverter 22. The control device 23 is further configured to convert the stator voltage U, V, W from the output voltage U_A. In doing so, the control device 23 receives the measurement data acquired by the at least one sensor 13 in order to adjust the stator voltage U, V, W based on the calculated excitation current. In further exemplary embodiments, the control device 23 is arranged outside the inverter 22.

[0045] Fig. 3 shows a schematic representation of an algorithm of an artificial intelligence applicable in the method according to the invention and the control device 23 according to the invention.

[0046] The Fig. The algorithm (AI model) schematically shown in Figure 3 for the artificial intelligence applied in the method described above includes the step of determining or reading out A1 an operating point. Subsequently, measurement data is determined A2 based on a reference device 30 described below for determining training data for the method described above or from a computer simulation based on a physical model of the separately excited electrical machine 10.

[0047] In the next step, physical quantities can be calculated from the measured data (A3). In the simplest case, the calculated physical quantities can be the calculation of a voltage from a measured current and a known resistance (U=RI), or a current density based on a known conductor cross-section, etc. These calculated physical quantities can then also be used as input data in the algorithm.

[0048] Finally, the input data is fed into the actual data-based model, which then determines the excitation current I_R A4. The artificial intelligence can contain at least a Gaussian process algorithm, a neural network, or a multi-layer perceptron algorithm. In the case of a Gaussian process algorithm, the uncertainty information can be used as a criterion to validate the demonstrated algorithm (AI model).

[0049] Fig. 4 shows a schematic representation of a method for determining training data for a method for estimating an excitation current I_R in a separately excited electrical machine 10 according to an embodiment of the present invention.

[0050] The reference device 30 comprises a Fig. 2. In addition, the reference device 30 comprises a telemetry sensor 31, which is arranged in the rotor 12 and is arranged and configured to detect the impressed excitation current I_R. A data processing device 32 connected to the telemetry sensor 31 is configured to operate the electric machine 10 over an operating range in order to determine reference excitation current data from the detected excitation current I_R and reference measurement data from the detected measurement data, and to provide training data for an artificial intelligence based on the determined reference measurement data and the reference excitation current data. The operating range of the electric machine 10 runs from idle to a maximum usable torque for the electric machine 10.To capture the entire operating range, the electric machine 10 of the reference device 30 is controlled accordingly by the inverter 22. An active learning method can be used to determine the reference measurement data and reference excitation current data, allowing relevant reference data to be determined more quickly.

[0051] Also in Fig. 4 shows two coils L1 and L2 arranged in the coupler 24, which are spaced apart from each other at a distance M. Thus, in this exemplary embodiment, a direct current I_dc is first generated from the output voltage U_A, which is converted by the second inverter 25 into an AC primary current that flows through the coil L1. An AC excitation current is induced at the coil L2 by inductive coupling, which AC excitation current is converted by a rectifier 12c into the excitation current I_R as direct current. The rectifier 12c for rectifying the excitation current I_R is designed as a diode 12c in this exemplary embodiment. In addition to the high rotor temperatures during operation of the electric machine 10, the diode 12c also generates a non-linear behavior of the excitation current I_R.

[0052] The control unit 23, which calculates the excitation current I_R based on the measurement data using artificial intelligence, is not absolutely necessary in the reference device 30, since the telemetry sensor 12d can measure this excitation current I_R directly. Nevertheless, the control unit 23 can calculate the excitation current I_R using artificial intelligence and the training data based on the remaining measurement data as input data and compare it directly with the measured value. The measurement data also include at least a terminal voltage, a rotational speed φ of the rotor 12, an active current, and a reactive current, as well as a temperature T_S, which is measured by a temperature sensor on the stator 11. Furthermore, the measurement data can contain the primary current I_P, which is transmitted from the second inverter 25 to the control unit 23.Furthermore, the measurement data may contain a direct voltage, in this case the output voltage U_A and / or a direct current I_dc, which in this case was generated by the output voltage U_A.

[0053] In Fig. 4 further shows that the electric machine 10, during operation, sets an axle 51, for example of a vehicle 100, in rotation via a transmission 50.

[0054] Fig. 5 shows a schematic representation of a system for estimating an excitation current I_R in a separately excited electrical machine 10 in a vehicle 100 according to an embodiment of the present invention.

[0055] The Fig. The system 200 shown in Figure 5 for estimating an excitation current in a separately excited electrical machine 10 in a vehicle comprises a vehicle 100 having a battery 14 for providing an output voltage. In the vehicle 100, a Fig. 2 is integrated. Furthermore, the system 200 comprises a device 20 Fig. 4 described reference device 30 with an electrical machine 10. The remaining components of the reference device 30 are in Fig. 5 not shown. In the Fig. In the system 200 shown in Figure 5, the data processing device 32 of the reference device 30 is configured to generate an update of the training data based on newly determined reference measurement data and reference excitation current data and to transmit it to the control device 23 of the device 20 in the vehicle 100 by means of a wireless connection 60.

[0056] The electric machine 10 in the vehicle 100 is now set by the inverter 22 or the control unit 23 to a predetermined torque and / or a predetermined speed of the rotor based on the excitation current I_R estimated with the updated or renewed reference current data.

[0057] The vehicle 100 can be included in a vehicle fleet consisting of a plurality of vehicles 100. Each vehicle 100 of this vehicle fleet contains a device 20 integrated into the respective vehicle 100. The data processing device 32 is designed to transmit the update to each vehicle 100 of the vehicle fleet. In preferred embodiments, the electric machine 10 of the reference device 30 is itself arranged in a vehicle 100. In this case, the data processing device 32 is designed to provide training data based on measurement data and excitation current data determined during operation of the vehicle 100. Thus, training data can be obtained under realistic conditions, which leads to more accurate estimates of the excitation current I_R.

[0058] Although the present invention has been fully described above using the preferred embodiment, it is not limited thereto but can be modified in many ways. QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] DE 11 2020 007 296 T5

[0003]

Claims

[1] Method for estimating an excitation current (I_R) in a separately excited electrical machine (10), comprising the steps: - providing (S1) an output voltage by a voltage source (14); - converting (S2), by an inverter (22), the output voltage (U_A) into a stator voltage (U, V, W) which is applied to a stator (11) of the electrical machine (10) for controlling a torque and / or a rotational speed (φ) of a rotor (12) of the electrical machine (10); - impressing (S3) an excitation current (I_R) into the rotor (12) by means of contactless energy transfer from a primary current generated from the output voltage; - detecting (S4), with at least one sensor (13), measurement data on the stator (11) for providing input data, wherein the measurement data includes at least a terminal voltage, the rotational speed (φ) of the rotor (12), an active current and a reactive current; and - Calculating (S5), with a processor (231) of a control device (23), the excitation current (I_R) by means of artificial intelligence based on the input data and training data. [2] Method according to claim 1, wherein the primary current is an AC primary current, wherein the contactless energy transfer is carried out by means of inductive coupling in a coupler (24), wherein an AC excitation voltage is induced in a rotor section (12b) of the coupler arranged in the rotor (12) by the AC primary current (I_P) in a primary section (24a) of the coupler (24), and wherein an AC excitation current (I_R) induced from the AC primary current (I_P) is rectified into the excitation current (I_R) by a rectifier (12c) in the rotor (12). [3] Method according to claim 1 or 2, wherein the at least one sensor (13) comprises a plurality of sensors (13), wherein from the plurality of sensors (13) at least one current sensor is arranged and designed to measure the active current and the reactive current in the inverter (22). [4] Method according to one of the preceding claims, wherein the artificial intelligence contains at least one Gaussian Process algorithm, a neural network, or a multi-layer perceptron algorithm. [5] Method according to one of the preceding claims, with the additional step: - Calculating a physical quantity from the measured data, where the input data includes the calculated physical quantities. [6] Method according to one of the preceding claims, wherein the control device (23) is arranged within the inverter (22) and is designed to convert the stator voltage (U, V, W) from the output voltage (U_A), wherein the control device (23) receives the measurement data acquired by the at least one sensor (13) in order to adjust the stator voltage (U, V, W) on the basis of the calculated excitation current. [7] Method according to one of the preceding claims, wherein the output current generated by the output voltage (U_A) is a direct current which is converted into the primary current (I_P) by a second inverter (25). [8] Method according to one of the preceding claims, wherein the measurement data further contain a temperature (T_Stator), the primary current, a direct voltage, in particular the output voltage (U_A), and / or a direct current (I_dc), which was generated in particular by the output voltage (U_A). [9] Method according to one of the preceding claims, wherein the training data are determined from reference measurement data which are detected by a telemetry sensor (12d), wherein the telemetry sensor (12d) is arranged and designed to detect the excitation current (I_R) of a rotor (11) of an electrical machine (10) of a reference device (30) induced by means of a contactless energy transfer, and / or the training data are determined by a computer simulation based on a physical model of the separately excited electrical machine (10). [10] The method of claim 9, further comprising the step of: - Updating the training data based on newly determined reference measurement data which were determined with the electric machine (10) of the reference device (30). [11] Method according to one of the preceding claims, further comprising: - controlling, with the inverter 22 or the control device (23), the electric machine (10) based on the calculated excitation current to set a predetermined speed (φ) and / or a predetermined torque. [12] Device (20) for a vehicle, comprising a voltage source (14) for providing an output voltage U_A; a separately excited electrical machine (10) having a stator (11) and a rotor (12), a coupler (24) which is designed to impress a primary current (I_P) generated by the output voltage into an excitation current (I_R) in the rotor (12) by contactless energy transfer, an inverter (22) of a torque and / or a rotational speed (φ) of a rotor (12) of the electric machine (10) by converting the output voltage (U_A) into a stator voltage (U, V, W); at least one sensor (13) for detecting measurement data on the stator (11) of the electrical machine (10), wherein the measurement data contain at least a terminal voltage, a rotational speed of the rotor, an active current and a reactive current, and a control device (23) which is connected to the electrical machine (10) and the at least one sensor (13), and which has a processor (231) and a non-volatile memory (232), wherein the memory (232) contains computer-readable instructions with which the processor (231) is able to calculate the excitation current (I_R) by means of artificial intelligence based on the measurement data as input data and training data. [13] Reference device (30) for determining training data for a method for estimating an excitation current in a separately excited electrical machine (10), in particular for a method according to one of claims 1 to 11, comprising a device (20) according to claim 12, a telemetry sensor (31) arranged in the rotor (12), which is arranged and designed to detect the impressed excitation current (I_R); a data processing device (32) connected to the telemetry sensor (31), which is designed to operate the electrical machine (10) over an operating range in order to determine reference excitation current data from the detected excitation current (I_R) and reference measurement data from the detected measurement data, and to provide training data for an artificial intelligence based on the determined reference measurement data and the reference excitation current data. [14] System (200) for estimating an excitation current in a separately excited electrical machine (10) in a vehicle (100), comprising: a vehicle (100) containing a battery (14) for providing an output voltage (U_A), a device (20) integrated in the vehicle (100) according to claim 12, and a reference device (30) according to claim 13, wherein the data processing device (32) of the reference device (30) is designed to generate an update of the training data based on newly determined reference measurement data and reference excitation current data and to transmit it to the control device (23) of the device (20). [15] The system (200) of claim 14, wherein the vehicle (100) is included in a vehicle fleet, each vehicle of the vehicle fleet including a device (20) according to claim 11 integrated in the respective vehicle (100), wherein the data processing device (35) is configured to transmit the update to each vehicle (100) of the vehicle fleet. [16] System (200) according to claim 14 or 15, wherein the electric machine (10) of the reference device (30) is arranged in a vehicle (100), wherein the data processing device (32) is designed to provide training data based on measurement data and excitation current data determined during operation of the vehicle (100).

Images