METHOD AND DEVICE FOR DETERMINING AT LEAST ONE ROAD PROPERTIES FOR AN ELECTRIC VEHICLE, AS WELL AS ELECTRIC VEHICLE

Description

[0001] The invention relates to a method and a device for determining at least one road surface property for an electric vehicle.

[0002] From US Patent 2016 / 0221581 A1, a method for classifying a road surface driven on by a vehicle is known. The method comprises receiving one or more electrical signals, each representing a vibration detected by a sensor mounted on the vehicle. The method further comprises identifying a pattern in the detected vibration represented by the one or more signals for at least one of the received electrical signals and matching the identified pattern to one of one or more known patterns, each known pattern corresponding to a specific road surface classification. The method further comprises classifying the road surface according to the road surface classification based on the known pattern that matches the identified pattern.A system comprising one or more vehicle-borne sensors configured to detect vibration and a pattern classification system for performing the procedure described above are also described.

[0003] From US Patent 2002 / 0162389 A1, a method for estimating a vehicle's driving condition is known. In this method, the initial vibration level of a section below the suspension of a vehicle, detected by a vibration sensor attached to that section, is frequency-converted using a frequency analyzer to obtain the frequency spectrum of the vibration level. An operation is performed on at least two vibration levels in different frequency bands of the obtained frequency spectrum using a vibration level calculator, and this calculated value is compared to a master curve representing the frequency spectrum of the vibration level stored in the vibration level memory to estimate the condition of a road surface and the driving condition of the vehicle.Furthermore, the running condition of each tire, including the tire pressure, is detected from the vibration level of the section below the vehicle's suspension to estimate the vehicle's running condition, thus providing a multifunctional sensing system for estimating the condition of a road surface or the running condition of a tire with a sensor.

[0004] A vehicle condition detection device is known from EP 2 933 161 A1. An unsprung condition detection component calculates a variation value, which is the magnitude of the difference between a detected angle output by a resolver rotary angle sensor for detecting the rotation angle of an in-wheel motor and an estimated motor rotation angle. The estimated angle can be calculated by adding an estimated motor rotation angle in one calculation cycle to a detected angle from the previous calculation cycle. If the variation value exceeds the road surface determination threshold, the unsprung condition detection component determines that the road on which the vehicle is traveling is rough. Consequently, the road surface can be determined using the rotary angle sensor.

[0005] From Bastian Weber, Positionsgeberlose Regelung von permanentmagneterexcitedn Synchronmaschinen bei kleinen Drehen mit überabprobender Stromerfassung, Hannover, Gottfried Wilhelm Leibniz Universität Hannover, Dissertation, 2018, 163 pp., https: / / doi.org / 10.15488 / 9140, a method for estimating a rotor position and / or a rotor speed is known.

[0006] The invention is based on the objective of creating a method and a device for determining at least one road surface property for an electric vehicle.

[0007] The problem is solved according to the invention by a method with the features of claim 1 and a device with the features of claim 9. Advantageous embodiments of the invention are set forth in the dependent claims.

[0008] According to the invention, a method for determining at least one road surface property for an electric vehicle is provided, wherein a wheel position and / or a wheel speed of at least one wheel of the electric vehicle is detected at high frequency by means of at least one wheel sensor, wherein a rotor position and / or a rotor speed of an electric machine coupled to the at least one wheel is detected at high frequency by means of a rotary encoder or estimated at high frequency without a rotary encoder, wherein, starting from the detected wheel position and / or wheel speed, a rotor position and / or rotor speed of the electric machine is estimated, wherein, from the estimated rotor position and / or the estimated rotor speed and the detected or encoder-free estimated rotor position and / or rotor speed, a rotor position error and / or rotor speed error is determined, wherein the determined rotor position error and / or the determined rotor speed error is evaluated over time.and wherein, based on an evaluation result, at least one road property is determined and provided as a road property signal.

[0009] Furthermore, a device for determining at least one road surface property for an electric vehicle is provided, comprising an evaluation unit, wherein the evaluation unit is configured to receive a wheel position and / or wheel speed of at least one wheel of the electric vehicle detected at high frequency by means of at least one wheel sensor, to receive a rotor position and / or a rotor speed of an electric machine coupled to the at least one wheel, detected by means of a rotary encoder or estimated without a rotary encoder, and to estimate a rotor position and / or rotor speed of the electric machine based on the detected wheel position and / or wheel speed, and to determine a rotor position error and / or rotor speed error from the estimated rotor position and / or the estimated rotor speed and the received detected or estimated rotor position and / or rotor speed.to evaluate the specified rotor position error and / or the specified rotor speed error over time, and based on an evaluation result, to determine at least one road surface property and provide it as a road surface property signal.

[0010] The method and the device make it possible to determine at least one road surface property via excitations transmitted from a surface or road surface through the wheels to the drive train of an electric vehicle. This is based on the idea that an electric motor in an electric vehicle is directly connected to the wheels via a transmission. In such a drive train, a non-uniform surface or road surface acts back on the electric motor. When all components are engaged, i.e., the transmission is not disengaged, the traction properties of a non-uniform surface, and thus the road surface properties, manifest themselves in a resulting drive train vibration.Since the electric motor generates significantly fewer of its own excitations compared to an internal combustion engine, the external excitations in the form of drivetrain vibrations can be much more easily attributed to excitations from a non-uniform road surface. This is achieved by high-frequency detection of the wheel position and / or wheel speed of at least one wheel of the electric vehicle using at least one wheel sensor. For example, sensors already used for an anti-lock braking system (ABS) on the wheels can be employed. Furthermore, the rotor position and / or rotor speed of an electric motor coupled to the at least one wheel is detected at high frequency using a rotary encoder or estimated at high frequency without a rotary encoder. Based on the detected wheel position and / or wheel speed, the rotor position and / or rotor speed of the electric motor is estimated.This is done in particular taking into account transmission elements arranged between the at least one wheel and the electric machine, such as a differential and / or a (reduction) gearbox. A rotor position error and / or a rotor speed error is determined from the estimated rotor position and / or the estimated rotor speed and the detected or encoder-free estimated rotor position and / or rotor speed. The determined rotor position error and / or the determined rotor speed error are evaluated over time. This is done in particular to identify vibrations. Based on an evaluation result of this analysis, at least one road surface property is determined and provided as a road surface property signal.Determining at least one road surface property is possible, in particular, through high-frequency detection of the wheel position and / or wheel speed and high-frequency detection or estimation of the rotor position and / or rotor speed. High frequency here means, in particular, that the detection or sampling frequency is significantly higher than the respective rotational speed, especially by a factor of 10 to 1000 times. High-frequency detection or sampling enables, in particular, high-resolution signal processing and signal enhancement through filtering, especially for noise reduction.

[0011] The method and the device have the advantage that at least one road surface property can usually be used without additional sensors, since wheel sensors are generally already present on the wheels of the electric vehicle.

[0012] A road surface property refers specifically to a characteristic of a subsurface, regardless of whether an artificial roadway is present or not. A road surface property can, for example, include at least one of the following features: a road surface, a road covering or subsurface material (fine / coarse asphalt, gravel, chippings, concrete slabs, sand, ice, etc.), as well as larger objects on the roadway (e.g., stones, etc.).

[0013] A wheel position is, in particular, a wheel angle. A rotor position is, in particular, an (electrical) rotor angle.

[0014] Parts of the device, in particular the evaluation unit, can be designed individually or collectively as a combination of hardware and software, for example, as program code executed on a microcontroller or microprocessor. However, it is also possible for parts to be designed individually or collectively as an application-specific integrated circuit (ASIC) or a field-programmable gate array (FPGA). The evaluation unit can also be part of a vehicle control system.

[0015] It may be provided that the wheel sensor(s) and / or other sensors on the drivetrain are part of the device.

[0016] It may be possible to transmit at least one specific road surface property to a vehicle control system. For example, this specific road surface property can be taken into account during traction control.

[0017] It is specifically intended that the evaluation of the high-frequency detected wheel position and / or wheel speed of at least one wheel and / or the high-frequency detected rotor position and / or rotor speed is performed using a field-programmable gate array (FPGA). In particular, it may be provided that the evaluation is performed using a system-on-chip (SoC) architecture that includes the field-programmable gate array. Furthermore, it may be provided that, as an alternative, the rotor position and / or rotor speed of the electric machine is also estimated using the field-programmable gate array within the framework of sensorless control. The rotor position here refers to the rotor angle. In particular, this estimation is based on high-frequency, i.e., oversampled, currents detected at the electric machine and / or at a drive inverter supplying it.Estimating the rotor position and / or rotor speed is described, for example, in Bastian Weber, Positionsgeberlose Regelung von permanentmagneterrangten Synchronmaschinen bei kleinen Drehen mit überabprobender Stromerfassung, Hannover, Gottfried Wilhelm Leibniz Universität Hannover, Dissertation, 2018, 163 pp., https: / / doi.org / 10.15488 / 9140.

[0018] Through the parallel signal processing of a field-programmable gate array, it can advantageously process high-frequency (especially ~MHz) sampled quantities. High-frequency sampling reduces sampling and aliasing effects as noise in the signal. Any noise components from the sensor can be digitally filtered in the field-programmable gate array, thus providing a very high signal-to-noise ratio for further digital processing. In comparison, a microcontroller (e.g., a DSP) is significantly less suitable for processing high-frequency signals, as the acquired values ​​of the DSP must be transported from the acquiring internal analog-to-digital converter (ADC) to the processor core and processed, which leads to delays in evaluation due to signal propagation delays and preprocessing.In a field-programmable gate array, the acquired values ​​of the ADC are immediately available to the logic units, i.e., in real time. Furthermore, field-programmable gate arrays can be used to evaluate cost-effective ADCs capable of generating a high-frequency digital data stream (bitstream), such as delta-sigma converters, without any additional measures.

[0019] The field-programmable gate array is, in particular, a field-programmable gate array that can be partially reconfigured at runtime. This makes the FPGA updateable. Reconfiguration is achieved primarily through partial reconfiguration, where only a portion of the gate array is reconfigured or reprogrammed at runtime, while other parts of the array retain their respective configurations or programming. Reconfiguration is typically performed using a configuration controller, which executes and monitors the process. Methods for automatic code generation, such as the "Xilinx System Generator for DSP," which can automate both source code modeling and functional analysis, can be used to reconfigure the at least one field-programmable gate array.The "System Generator for DSP" enables, for example, the modeling of FPGA functions in a Matlab Simulink environment, allowing their properties to be tested against other Simulink function blocks. Automatic code generation ensures that the field-programmable gate array behaves identically to the Simulink model.

[0020] A configuration of a field-programmable gate array includes, in particular, instructions with which the field-programmable gate array can be programmed in such a way that a certain functionality for data processing can be provided by means of the programmed (or configured or reconfigured) part of the field-programmable gate array.

[0021] In principle, the sensorless estimation of the rotor position and / or rotor speed of the electric machine can also be performed using an ASIC. The evaluation and determination of at least one road surface property can also be carried out using a dedicated ASIC.

[0022] In one embodiment, the drive-side differential speed of a differential is additionally detected at high frequency using a speed sensor. Based on detected wheel positions and / or wheel speeds of wheels of the electric vehicle connected via the differential, the drive-side differential speed of the differential is estimated. A differential speed error is then determined from the estimated drive-side differential speed and the detected drive-side differential speed and taken into account when evaluating and determining at least one road surface property. This allows vibrations in the drivetrain to be separated according to their respective components. In particular, this allows a vibration caused by the differential to be separated from a vibration observed at the electric motor.Several errors are then available, the temporal evolution of which can be evaluated to determine at least one road surface property. In a simple case, the estimated drive-side differential speed can be determined by averaging the recorded speeds of the two wheels connected to the differential.

[0023] In one embodiment, an amplitude spectrum and / or a phase spectrum of the time-dependent behavior of the specified rotor position error and / or the specified rotor speed error, and optionally also the specified differential speed error, is determined. The amplitude spectrum and / or the phase spectrum are then evaluated. This allows for the particularly efficient identification of excitations caused by the road surface and transmitted into the drivetrain. For example, a short-time Fourier transform (STFT) can be performed. This allows individual frequency components to be identified and evaluated over time.Since harmonics caused by road surface properties usually do not result in an integer multiple of the drivetrain rotational speeds, they can be more easily separated from self-excited drivetrain vibrations, provided that a window function of the short-time Fourier transform encompasses a multiple of the period or the rotational time of individual drivetrain components (wheels, differential, transmission, electric motor, etc.). Self-excited drivetrain vibrations include, for example, excitations from the tooth force level (multiples of the gear tooth count) or excitations from the electric motor (harmonics from an induction matrix of the electric motor / a reluctance torque, harmonics from flux linkage of the rotor, dead-time effect of an inverter feed, etc.).

[0024] In one embodiment, the evaluation result is compared with a database in which evaluation results are linked to road surface properties, and the at least one road surface property is determined based on a comparison result. This allows the at least one road surface property to be determined based on empirically and / or simulation-determined relationships between the temporal behavior of the rotor position error and / or the rotor speed error and / or the differential speed error with various road surface properties. In particular, specific patterns within a specific amplitude spectrum and / or phase spectrum of the temporal profiles can be recognized and evaluated in order to identify a similar pattern in the database by comparison and to retrieve and provide the associated road surface property(ies) from the database.

[0025] In one embodiment, the evaluation result for determining at least one road surface property is fed to at least one trained machine learning method. This trained machine learning method is configured and trained to estimate the at least one road surface property based on the evaluation result. During a training phase, the machine learning method is trained using a training dataset. This dataset contains training data with associated fundamental truths. The training data includes, in particular, time profiles of a wheel position error and / or a wheel speed error and / or a differential speed error, each linked to at least one road surface property. The training data and the associated fundamental truths were determined, in particular, empirically.In principle, the training data can also be partially or completely generated through simulation. The training data is fed into the machine learning program as input, and the machine learning program estimates at least one road property as output. The estimated road property is compared to the baseline, and based on the comparison result, parameters of the machine learning program are adjusted. This process is repeated until the machine learning program achieves sufficient accuracy. After the training phase, the trained machine learning program is made available. For this purpose, a structural description and parameters of the trained machine learning program are loaded into a memory of the evaluation unit, for example, for use. The machine learning program is, in particular, an artificial neural network, specifically a convolutional neural network (CNN).

[0026] In one embodiment, the temporal evolution of the specified rotor position error and / or the specified rotor speed error and / or the specified differential speed error is filtered to isolate self-excited drivetrain vibrations. This allows for better identification of external excitations caused by road surface characteristics. For this purpose, current drivetrain parameters, such as transmission characteristics and / or transmission parameters, etc., can be evaluated to identify frequencies and / or frequency ranges of the self-excited drivetrain vibrations. The identified frequencies and / or frequency ranges are then filtered (low-pass, band-pass, high-pass, etc.) to reduce their contribution in the subsequent evaluation.

[0027] In one embodiment, the measures are carried out for at least two separately driven axles of the electric vehicle, with the at least one road surface property being determined taking into account the respective evaluation results. This allows for a more accurate determination of the at least one road surface property, as information from more than one drive axle is available.

[0028] In one embodiment, a warning message is generated and displayed depending on at least one specific road surface property. For example, a visual, audible, and / or haptic warning signal may be displayed to the driver if at least one specific road surface property impairs the safety of the electric vehicle, taking into account the current driving style.

[0029] Further features regarding the design of the device are derived from the description of embodiments of the method. The advantages of the device are the same in each case as in the embodiments of the method.

[0030] Furthermore, an electric vehicle will be created, comprising at least one device according to one of the described embodiments. In principle, the electric vehicle can also be a hybrid vehicle, provided there are periods in which it is operated purely electrically.

[0031] The invention is explained in more detail below with reference to preferred embodiments and the figures. These show: Fig. 1 a schematic representation of an embodiment of the device for determining at least one road surface property for an electric vehicle; Fig. 2 a schematic representation to illustrate an example for determining the rotor speed error; Fig. 3 a schematic representation to illustrate the generation of excitations by road surface properties or properties of a subgrade; Fig. 4 a schematic representation to illustrate an embodiment of the method and the device; Fig. 5 a further schematic representation to illustrate the embodiment of the method and the device; Fig. 6 a schematic representation to illustrate a short-time Fourier transform; Fig. 7 a schematic flowchart of an embodiment of the method for determining at least one road surface property for an electric vehicle.

[0032] In Fig. 1Figure 1 shows a schematic representation of an embodiment of the device 1 for determining at least one road surface property 40 for an electric vehicle 50. The device 1 is arranged in the electric vehicle 50.

[0033] The electric vehicle 50 comprises an electric machine 52 on a rear drive axle 51, which is connected to the wheels 54 of the rear axle 51 via a (reduction) gearbox 53. The electric machine 52 is controlled by means of a drive inverter 55.

[0034] The electric vehicle 50 also has a wheel sensor 56 on each of the wheels 54 of the rear axle 51, which, for the sake of clarity, is only shown for one wheel 54. The wheel sensor 56 is used to determine the wheel position. ϕ 1 and / or wheel speed n The position of wheel 54 is detected at high frequency and transmitted as a corresponding signal to the device 1.

[0035] A rotor position ϕ EM and / or a rotor speed name The rotor position is detected at high frequency by means of a rotary encoder 57 on the electric machine 52 and transmitted as a corresponding signal to the device 1. Alternatively, the rotor position can be detected. ϕ EM and / or the rotor speed name also appreciated, especially in the context of encoderless control of the electric machine 52, as described, for example, in Bastian Weber, Positionsgeberlose Regelung von permanentmagneterrangten Synchronmaschinen bei kleinen Drehen mit überabprobender Stromerfassung, Hannover, Gottfried Wilhelm Leibniz Universität Hannover, Dissertation, 2018.

[0036] The device 1 comprises an evaluation unit 2. The evaluation unit 2 is configured to evaluate the wheel position detected at high frequency by means of the wheel sensor system 56. ϕ 1 and / or wheel speed nThe evaluation unit 2 receives the rotor position detected by the rotary encoder 57 or, alternatively, the rotor position estimated without a rotary encoder. ϕ EM and / or the rotor speed detected by means of the rotary encoder 57 or the rotor speed estimated without a rotary encoder name the electric machine 52 coupled to the wheel 54.

[0037] Starting from the received wheel position φ 1 and / or wheel speed n 1. The evaluation unit 2 estimates a rotor position. φ̂ EM and / or rotor speed name of the electric machine 52.

[0038] From the estimated rotor position φ̂ EM and / or the estimated rotor speed name and the received, detected, or encoder-free estimated rotor position ϕ EM and / or rotor speed name The evaluation unit 2 determines a rotor position error. and φEM and / or rotor speed errors and nEM .

[0039] Evaluation unit 2 evaluates the determined rotor position error. and φEM and / or the specific rotor speed error and nEM The system analyzes the behavior over time and, based on an evaluation result, determines at least one road surface property 40. For this purpose, patterns can be recognized over time and compared with patterns that occur for known road surface properties to determine the at least one road surface property. The determined at least one road surface property 40 is provided as a road surface property signal 41, for example, in the form of an analog or digital signal, such as a digital data packet. The at least one road surface property 40 is then fed, for example, to a vehicle control system 60, which, for example, performs traction control.

[0040] It may be provided that a warning message 42 is generated and issued depending on at least one specific road surface characteristic 40. This occurs, for example, if the at least one road surface characteristic 40 is intended to encourage a driver of the electric vehicle 50 to adopt a less aggressive or less sporty driving style.

[0041] Evaluation unit 2 is provided, at least in part, by means of a field-programmable gate array (not shown). Specifically, the field-programmable gate array is part of a system-on-chip (SoC) architecture. This enables fast processing and evaluation of the high-frequency acquired and / or estimated quantities.

[0042] In the Fig. 2 is determining the rotor speed error and nEM This is illustrated schematically using an example. Starting with the measured wheel speed. n1 is achieved using a gear ratio ü, which is a gear ratio of the gearbox 53 ( Fig. 1 ) describes a rotor speed name The electric machine 52 was estimated. By taking the difference between the rotor speed thus estimated name and the recorded rotor speed name will the rotor speed error and nEM The rotor position error is determined in a fundamentally analogous manner.

[0043] It may be provided that a drive-side differential speed n T a differential 58 ( Fig. 1 ) is detected at high frequency by means of a speed sensor 59, starting from detected wheel positions ϕ 1 , ϕ 2 and / or wheel speeds n 1 , n 2 of the wheels 54 of the electric vehicle 50 connected via the differential 58 a drive-side differential speed n̂ Tof the differential 58 is estimated, from the estimated drive-side differential speed n̂ T and the recorded drive-side differential speed n T a differential speed error and nT determined (cf.) Figure 4 and 5 ) and is taken into account when evaluating and determining at least one roadway property 40.

[0044] It can be provided that the measures are carried out for at least two separately driven drive axles 51, 71 of the electric vehicle 50, whereby at least one road surface property 40 is determined taking into account the respective evaluation results. The procedure for drive axle 71 is analogous to the procedure described above with reference to drive axle 51. The in the Fig. 1 The components shown with reference to the drive axis 71 are therefore not separately marked with reference symbols.

[0045] In Fig. 3Figure 100 is a schematic representation illustrating the generation of excitations due to road surface properties or the properties of a substrate. For this purpose, a wheel 54 with a manufacturing radius is shown. R Modeled at 0. Due to weight forces, wheel 54 is flattened on a mounting surface, resulting in a stationary wheel radius. R started This results in the dynamic wheel radius. R dyn This results from rolling tests.

[0046] A drive torque acts on wheel 54. M Antr Due to unevenness on the road surface 100 or the subsoil (symbolized here by stones 101), reaction forces occur when rolling in a direction of travel 102. F ⇀ reak at wheel 54 and a reaction moment results M reak If the excitation on wheel 54 by the road surface 100 is small, then the reaction torque is also small. M reactsmall. The depicted stones 101 can be significantly smaller; for example, they can be formed as fine gravel. Since the device 1 ( Fig. 1 However, due to the high-frequency sampling, it has a high resolution, allowing even small excitations to be detected. Since the reaction time M react Since the reaction cannot be directly measured, the response of the drivetrain is evaluated instead. The reaction torque M react This causes a harmonic in the drive train, which is evaluated using the method described in this disclosure.

[0047] The Figure 4 and 5 illustrate an embodiment of the method and device 1 ( Fig. 1 ), in which it is provided that a drive-side differential speed n T of a differential 58 (rotary gear carrier) by means of a speed sensor 59 ( Fig. 1 ) is detected at high frequency. Based on detected wheel speeds. n 1 ,n 2 of the wheels 54 of the electric vehicle, which are connected to each other via the differential 58, are given a drive-side differential speed. n̂ T of the differential 58 estimated, from the estimated drive-side differential speed n̂ T and the recorded drive-side differential speed n T a differential speed error and nT determined and taken into account when evaluating and determining at least one roadway property.

[0048] This is in Fig. 5 clarifies. The recorded wheel speeds n 1 , n 2 are used to determine the drive-side differential speed n̂ T to estimate, in the example shown, using the equation: n ^ T = 1 2 n 1 + n 2

[0049] The differential speed error is then calculated. and nT determined by the difference between the estimated drive-side differential speed n̂ Tand the recorded drive-side differential speed n T is formed.

[0050] From the recorded drive-side differential speed n T is achieved using a transmission ratio ü, which is a transmission ratio of the gearbox 53 ( Fig. 4 ) describes a rotor speed name the electric machine 52 ( Fig. 4 ) estimated. By calculating the difference between the rotor speed estimated in this way. name and the recorded rotor speed name will the rotor speed error and nEM certainly.

[0051] The specific rotor speed error and nEM The data is evaluated over time, whereby, based on an evaluation result, at least one roadway property is determined and provided as a roadway property signal.

[0052] Additionally, in this embodiment the differential speed error is also taken into account. and nTevaluated over time, whereby the evaluation result is taken into account when determining at least one roadway property.

[0053] In the Fig. 6 is a schematic representation to illustrate an embodiment of the method and the device 1 ( Fig. 1 ) shown. In the embodiment, which is otherwise like that shown in the Fig. 1 In the embodiment shown, it is provided that an amplitude spectrum and / or a phase spectrum of the time-dependent behavior of the specified rotor position error and / or the specified rotor speed error and / or the specified differential speed error is determined, wherein the amplitude spectrum and / or the phase spectrum are evaluated for analysis. For this purpose, a short-time Fourier transform is used in particular, which in the continuous case can be mathematically formulated as follows: X z ω = ∫ − ∞ ∞ x t w t − z e − iωt dt where w(·) denotes a window function with a shift time z. The shift time z is chosen in particular such that it is a multiple of a revolution period of a wheel 54 ( Fig. 1 ) is, for example, 10, 50, or 100 times larger than the orbital period. In the discrete case, which is used in this embodiment due to the digitized, high-frequency acquired or estimated measured values, the short-time Fourier transform is mathematically as follows: X m ω = ∑ n = − ∞ ∞ x n w n − m e − iωn

[0054] In the Fig. 6 The signal above is a discrete signal for clarification. x ( n ) over the individual discrete time points n shown. Among them is the short-time Fourier transform. X ( ω ) in the form of section-wise amplitude spectra 30 of the signal x ( n ) above the angular frequency ωThis is shown. If the signal is zero, then the transformed signal is also zero. Depending on the frequency components in the signal x ( n The amplitude spectrum of each component changes accordingly. The representation is merely an example for illustrative purposes. In a real-world evaluation, the signal may... x ( n ) and the short-time Fourier transform X ( ω ) be significantly more complex.

[0055] It may be provided that the evaluation result, in particular an amplitude spectrum and / or phase spectrum, is compared with a database in which evaluation results are linked to road surface properties 40, wherein at least one road surface property 40 is determined based on a comparison result. The database was, for example, empirically determined and is, for example, stored in a memory of the evaluation unit 2 ( Fig. 1) stored.

[0056] It may further be provided that the evaluation result, in particular an amplitude spectrum and / or phase spectrum, is fed to at least one trained machine learning method to determine the at least one roadway property 40, wherein the trained machine learning method is configured and trained to estimate the at least one roadway property 40 based on the evaluation result. The machine learning method is in particular an artificial neural network, in particular a convolutional network. The trained machine learning method is used by means of the evaluation device 2 ( Fig. 1 ) provided.

[0057] It may be provided that the time course of the specific rotor position error is and φEM ( Fig. 1 ) and / or the specific rotor speed error and nEM ( Fig. 1 ) and / or the specific differential speed error and nT ( Fig. 1) is filtered to remove self-excited drivetrain vibrations. For this purpose, frequencies and / or frequency ranges in the amplitude spectrum 30 ( Fig. 6 Data from these frequency ranges, where self-excited drivetrain vibrations are known to occur, are not considered during evaluation. For example, information from these frequencies or frequency ranges would not be fed into a machine learning process as input data; only information from the remaining frequency ranges would be used. Similarly, when comparing data with a database, information from these frequency ranges would not be taken into account.

[0058] In Fig. 7Figure 1 shows a schematic flowchart of an embodiment of the method for determining at least one road surface property for an electric vehicle. The method is carried out in particular by means of an embodiment of the device as described in the following. Fig. 1 shown.

[0059] In measure 200, the wheel position and / or wheel speed of at least one wheel of the electric vehicle is recorded at high frequency using at least one wheel sensor.

[0060] In measure 201, the rotor position and / or rotor speed of an electric machine coupled with at least one wheel is detected at high frequency using a rotary encoder or estimated at high frequency without a rotary encoder.

[0061] In measure 202, the rotor position and / or rotor speed of the electric machine is estimated using an evaluation device, based on the recorded wheel position and / or wheel speed.

[0062] In measure 203, a rotor position error and / or rotor speed error is determined from the estimated rotor position and / or the estimated rotor speed and the detected or encoder-free estimated rotor position and / or rotor speed.

[0063] In measure 204, the specific rotor position error and / or the specific rotor speed error will be evaluated over time.

[0064] Measure 204 may include the determination of an amplitude spectrum and / or a phase spectrum of the time-dependent behavior of the specified rotor position error and / or the specified rotor speed error, and optionally also of the specified differential speed error, whereby the amplitude spectrum and / or the phase spectrum are evaluated for analysis. For example, individual frequencies or frequency ranges can be searched for signals and / or patterns, and / or an amplitude and / or a phase at specific frequencies and / or in specific frequency ranges can be evaluated.

[0065] Furthermore, measure 204 may provide for filtering the time course of the specified rotor position error and / or the specified rotor speed error and / or the specified differential speed error in order to filter out self-excited drivetrain vibrations. The evaluation result will then specifically no longer include the filtered-out components.

[0066] Measure 205 stipulates that, based on an evaluation result, at least one road surface property is determined and made available as a road surface property signal. The road surface property signal can be implemented in either an analog or digital form, for example, as a digital data packet describing the at least one road surface property.

[0067] Measure 205 may include a comparison of the evaluation result with a database containing evaluation results linked to road surface properties, whereby at least one road surface property is determined based on a comparison result. For example, the characteristics of amplitudes and / or phases across the frequency spectrum can be compared with entries, such as patterns, in the database to find matches and derive at least one road surface property from them.

[0068] Furthermore, measure 205 may stipulate that the evaluation result for determining at least one road surface property is fed to at least one trained machine learning method, whereby the trained machine learning method is configured and trained to estimate the at least one road surface property based on the evaluation result. In particular, the trained machine learning method, for example, a trained artificial neural network, is provided with an amplitude spectrum and / or a phase spectrum as input data. The trained machine learning method then estimates the at least one road surface property based on this. The trained machine learning method can, in particular, recognize patterns and / or features in an evaluation result, for example, an amplitude spectrum and / or phase spectrum, and estimate the at least one road surface property based on the recognized patterns and / or features.

[0069] In measure 206, it may be provided that a warning message is generated and issued depending on at least one specific roadway property.

[0070] Subsequently, measures 200 to 206 can be repeated so that at least one current roadway property can be continuously determined and provided. Contact checklist

[0071] 1 Device 2 Evaluation unit 30 Amplitude spectrum 40 Road surface property 41 Road surface property signal 42 Warning message 50 Electric vehicle 51 Drive axle 52 Electric machine 53 Transmission 54 Wheel 55 Drive inverter 56 Wheel sensor 57 Rotary encoder 58 Differential 59 Speed ​​sensor 60 Vehicle control 71 Drive axle 100 Roadway 101 Stone 200-205 Measures and φEM Rotor position error and nEM Rotor speed error and nT Differential speed error F reak ⇀ Reaction forces ϕ 1 Wheel bearing ϕ 2 Wheel bearing ϕ EM Rotor position φ̂ EM estimated rotor position M Antr Drive torque M react reaction moment n 1 Wheel speed n 2 Wheel speed name Rotor speed n T Differential speed name estimated rotor speed n̂ T estimated drive-side differential speed R 0 manufacturing radius (wheel) R started stationary wheel radius R dyn dynamic wheel radius ü gear ratio (transmission) w (·) window function x(n) signal X ( ω ) Short-term Fourier transform

Claims

1. Method for determining at least one roadway property (40) for an electric vehicle (50), wherein a wheel position (φ1, φ2) and / or a wheel speed (n1, n2) of at least one wheel (54) of the electric vehicle (50) is detected at a high frequency by means of at least one wheel sensor system (56), wherein a rotor position (φEM) and / or a rotor speed (nEM) of an electric machine (52) coupled to the at least one wheel (54) is detected at a high frequency by means of a rotary encoder (57) or estimated at a high frequency without a rotary encoder, wherein a rotor position and / or rotor speed (n̂EM) of the electric machine (52) is estimated based on the detected wheel position (φ1, φ2) and / or wheel speed (n1, n2), wherein a rotor position error (eφEM) and / or a rotor speed error (enEM) is determined from the estimated rotor position and / or the estimated rotor speed (n̂EM) and from the rotor position (φEM) and / or rotor speed (nEM) which has been detected or has been estimated without a rotary encoder, characterized in that the determined rotor position error (eφEM) and / or the determined rotor speed error (enEM) are evaluated over time, and wherein, based on an evaluation result, at least one roadway property (40) is determined and provided as a roadway property signal (41).

2. Method according to claim 1, characterized in that a drive-side differential speed (nT) of a differential (58) is detected at a high frequency by means of a speed sensor (59), wherein a drive-side differential speed (n̂T) of the differential (58) is estimated based on the detected wheel positions (φ1, φ2) and / or wheel speeds (n1, n2) of the wheels (54) of the electric vehicle (50) which are connected to each other via the differential (58), wherein a differential speed error (enT) is determined from the estimated drive-side differential speed (nT) and the detected drive-side differential speed (nT), and taken into account when evaluating and determining the at least one roadway property (40).

3. Method according to claim 1 or claim 2, characterized in that i) an amplitude spectrum (30) and / or a phase spectrum of the time curves of the determined rotor position error (eφEM) and / or the determined rotor speed error (enEM) is determined, or ii) an amplitude spectrum (30) and / or a phase spectrum of the time curves of the determined rotor position error (eφEM) and / or the determined rotor speed error (enEM) and the determined differential speed error (enT) is determined; wherein the amplitude spectrum (30) and / or the phase spectrum are evaluated for the evaluation.

4. Method according to any of the preceding claims, characterized in that the evaluation result is compared with a database in which evaluation results linked to roadway properties (40) are stored, wherein the at least one roadway property (40) is determined based on a comparison result.

5. Method according to any of the preceding claims, characterized in that the evaluation result for determining the at least one roadway property (40) is fed to at least one trained machine learning method, wherein the trained machine learning method is designed and trained to estimate the at least one roadway property (40) based on the evaluation result.

6. Method according to any of the preceding claims, characterized in that the time curve of the determined rotor position error (eφEM) and / or the determined rotor speed error (enEM) and / or the determined differential speed error (enT) is filtered to filter out self-excited drivetrain vibrations.

7. Method according to any of the preceding claims, characterized in that the measures are carried out for at least two separately driven drive axles (51,71) of the electric vehicle (50), wherein the at least one roadway property (40) is determined taking into account the respective evaluation results.

8. Method according to any of the preceding claims, characterized in that a warning message (42) is generated and issued depending on the determined at least one roadway property (40).

9. Apparatus (1) for determining at least one roadway property (40) for an electric vehicle (50), comprising an evaluation device (2), wherein the evaluation device (2) is designed to receive a wheel position (φ1, φ2) and / or a wheel speed (n1, n2) of at least one wheel (54) of the electric vehicle (50), which wheel position and wheel speed are detected at a high frequency by means of at least one wheel sensor system (56), to receive a rotor position (φEM) detected by means of a rotary encoder (57) or estimated without a rotary encoder and / or a rotor speed (nEM) detected by means of the rotary encoder (57) or estimated without the rotary encoder, which rotor position and rotor speed are of an electric machine (52) coupled to the at least one wheel (54), and to estimate a rotor position (ϕ̂EM) and / or rotor speed (nEM) of the electric machine (52) based on the detected wheel position (φ1, φ2) and / or wheel speed (n1, n2), to determine a rotor position error (eφEM) and / or rotor speed error (enEM) from the estimated rotor position (ϕEM) and / or the estimated rotor speed (nEM) and from the received rotor position (φEM) and / or rotor speed (nEM) which has been detected or been estimated without a rotary encoder, characterized in that the evaluation device is further designed to evaluate the determined rotor position error (eφEM) and / or the determined rotor speed error (enEM) over time, and to determine at least one roadway property (40) based on an evaluation result and to provide it as a roadway property signal (41).

10. Electric vehicle (50), comprising at least one apparatus (1) according to claim 9.