Method for controlling an electric axle drive in a vehicle that is at least partially electrified in the event of a sensor defect or failure

Abstract

The present invention relates to a method (100) for controlling an electric axle drive (40) in an at least partially electrified vehicle (50), wherein the electric axle drive (40) comprises an electric drive unit (42) and a power output stage (46), the method (100) comprising: receiving a state representation (12) relating to one or more state parameters of the electric drive unit (42); generating an estimation result (14) of the state parameter or the multiple state parameters from the state representation (12) by means of an observer module (30); receiving a detection result (16) relating to the state parameter or the multiple state parameters, wherein the detection result (16) is obtained by means of one or more sensor devices; obtaining a fault pattern (18) of the detection result (16); and adapting the observer module (30) based on the fault pattern (18).Determining a final result (34) of the state parameter or several state parameters based on the estimation result (14) and the acquisition result (16) using the adapted observer module (30), wherein the final result (34) is intended to provide a control signal for controlling the power output stage (46).

Description

[0001] The invention relates to a method for controlling an electric axle drive in an at least partially electrified vehicle, a corresponding device or control unit, a corresponding electric axle drive with the device, an at least partially electrified vehicle with the electric axle drive, and a corresponding computer-readable storage medium.

[0002] In the prior art, electrified vehicles or electric vehicles are known which are driven exclusively or partially by an electric drive unit (in particular an electric machine or electric motor) as the drive system. To supply the electric drive unit of such electric vehicles with electrical energy, the electric vehicles include electrical energy storage devices, in particular rechargeable electric batteries or secondary batteries. These batteries are designed as DC voltage sources, but the electric drive units generally require AC voltage. Therefore, a power output stage, usually referred to as a so-called...The DC / AC inverter (inverter) is configured with semiconductor-based power electronics and is connected to convert the DC input voltage into an AC output voltage to power the electric drive unit.

[0003] These inverters use half-bridges to establish predetermined voltages on the strings or phases of the electric drive unit. In each half-bridge, two semiconductor-based power switches (high-side and low-side) are connected in series between different potentials of a DC link voltage. The corresponding phase of the electric drive unit is contacted at the point where the two power switches are connected. By alternately opening and closing the two power switches with a predetermined duty cycle, such as pulse-width modulation (PWM), the desired phase voltage, particularly a sinusoidal one, is established. In this way, the AC output voltage, or phase-shifted phase voltages, can be generated from the DC input voltage of the battery on the strings of the electric drive unit.These phase voltages generate a rotating stator magnetic field, which interacts with the rotor's own magnetic field and thus drives the rotor to rotate.

[0004] The power output stage or inverter is controlled based on an impressed control signal, which provides the predetermined duty cycle, in particular pulse width modulation. Various approaches for generating such control signals are known in the prior art. However, these known approaches require a comparatively large number of input variables, such as sensor data, which are generated by a corresponding number of sensors when acquiring state parameters of the electric drive unit. This leads to a relatively high computational effort and, at the same time, an increased probability of errors in the control signal to be generated.

[0005] Furthermore, a defect or even failure of at least one part of the sensor system often leads to a defect or failure of the entire drive system, necessitating complex and costly compensatory measures to resume smooth operation. A gradual transition from the failure state to normal operation of the drive system, achieved through compensatory measures and without abrupt fluctuations in the state parameters, is either impossible or only possible to a limited extent. The faulty or defective sensor data can generally only be compensated for by downstream correction or adaptation functions, which are computationally intensive and prone to errors.

[0006] It is an object of the invention to provide a method for controlling an electric axle drive in an at least partially electrified vehicle, which makes it possible to overcome the aforementioned disadvantages at least partially.

[0007] This problem is solved according to the invention by the method, the device, the electric axle drive with the device, the at least partially electrified vehicle with the electric axle drive, and the computer-readable storage medium according to the independent claims. Advantageous embodiments and further developments of the invention are described in the dependent claims.

[0008] The present invention relates in a first aspect to a method and a corresponding device for controlling an electric axle drive in a vehicle that is at least partially electrified.

[0009] The electrified vehicle (electric vehicle) can be a purely electric vehicle or a hybrid vehicle. The vehicle's electric axle drive comprises an electric drive unit, which is preferably designed as an electric machine (E-machine), for example, a synchronous machine (PSM / FESM / EESM) or an asynchronous machine (ASM). The vehicle also includes a DC power supply, which is preferably designed as a traction battery (e.g., lithium battery, lithium-ion battery, or fuel cell battery) with a predefined nominal voltage as the DC input voltage. This could, for example, be a high-voltage battery (HV battery) with a nominal voltage of 400V or 800V.

[0010] To power the electric drive unit, which (especially its stator) has several coil windings, the electric axle drive includes a power output stage comprising a DC / AC inverter. The inverter is connected between the drive battery, preferably between a DC link of the electric axle drive, and the electric drive unit. Thus, the inverter is configured to convert the DC input voltage provided by the DC power supply into an AC output voltage, which powers the electric drive unit for torque generation. The DC link generates a DC link voltage based on the DC power supply or the drive battery. The DC link may comprise a single capacitor or an array of several capacitors.

[0011] The inverter comprises a corresponding phase unit for each coil winding of the stator of the associated electric drive unit. In a multi-phase electric axle drive with several phases, the inverter therefore comprises several phase units. Each phase unit includes several semiconductor-based power switches that form a half-bridge. MOSFETs, IGBTs, or other packages can be selected for the power switches. Silicon or a wide-bandgap semiconductor (WBS) such as silicon carbide (SiC) or gallium nitride (GaN) can be used as the semiconductor material for the power switches.In each half-bridge, a high-side device consisting of one or more parallel-connected circuit breakers (high-side circuit breakers) and a low-side device consisting of one or more parallel-connected circuit breakers (low-side circuit breakers) are connected in series between different potentials of the DC link voltage. The corresponding phase of the electrical drive unit is contacted at the point where the high-side and low-side devices are connected. When the high-side and low-side circuit breakers are alternately opened and closed by means of control signals with a predetermined duty cycle, the desired phase voltage is established at the respective phase, which is then supplied to the corresponding coil winding.In this way, pulse width modulation (PWM) can be used to generate phase voltages that vary sinusoidally over time at the individual phases of the electrical drive unit, which together form the respective AC output voltage.

[0012] The electric drive unit is therefore energized and driven based on the control of the power output stage or the inverter. This is preferably achieved using field-oriented control (FOR), which enables the decoupling of the field- and torque-generating components, also referred to as the d- and q-components, respectively. Here, state parameters such as stator and rotor parameters, in particular the phase currents of the stator coil windings and the rotor speed, angular position, angular velocity, and angular acceleration of the rotor, are continuously acquired, preferably in real time. The phase currents are transformed into a rotor-fixed two-axis coordinate system (dq-coordinate system) to obtain a d-current component and a q-current component. These stator and rotor parameters are used as control variables of the FOR control loop to generate the control signals for driving the power switches.

[0013] The inventive method for controlling an electric axle drive comprises a process step of receiving a state representation relating to several state parameters of the electric drive unit. The state representation preferably describes several state parameters of the electric drive unit by means of a mathematical representation (e.g., by means of a state-space representation), which preferably include one or more rotor state variables such as an angular position, an angular velocity, an angular acceleration, and a rotor speed, and one or more stator state variables such as one or more phase currents, a d-current component, and / or a q-current component. Depending on the state parameters, the state representation describes an operating state of the electric drive unit.

[0014] The state representation preferably comprises one or more systems of equations, more preferably differential systems of equations. The state representation can, in particular, comprise a first system of equations for the d-current component and the q-current component, comprising the following equations: x1,k=x1,k−1⋅(1−Rs⋅TsLd)+x2,k−1⋅Lq⋅Ts⋅ωLd+u1⋅TsLd x2,k=x1,k−1⋅(−Ld⋅Ts⋅ωLq)+x2,k−1⋅(1−Rs⋅TsLq)+u2⋅TsLq−κ⋅Ts⋅ωLq where the variable x1 is the d-current component, the variable x2 is the q-current component, R s a stator resistor, T s a sampling interval of the device or control unit executing the method according to the invention, L d a stator inductance with respect to the longitudinal axis of the electric drive unit, L qa stator inductance with respect to the transverse axis of the electric drive unit, κ a pole flux density, ω an angular velocity of the rotor of the electric drive unit, u1 a d-voltage component and u2 a q-voltage component of the rotor-fixed biaxial coordinate system, k or k-1 a number of calculation steps (or calculation steps) of the sampling processes (or calculation steps). The d-voltage component and / or the q-voltage component can preferably be provided from the FOR control loop, for example, each as a control variable of the FOR control loop.

[0015] The state representation can alternatively include a second system of equations for the rotor state variables angular position, angular velocity and angular acceleration, comprising the following equations: x3,k=x3,k−1+x4,k−1⋅Ts+x5,k−1⋅Ts22 x4,k=x4,k−1+x5,k−1⋅Ts x5,k=x5,k−1 where the variable x3 represents an angular position of the rotor, the variable x4 the angular velocity of the rotor, the variable x5 an angular acceleration of the rotor, and k or k-1 a number of calculation steps of the sampling processes.

[0016] According to one embodiment, the second system of equations, which comprises three equations with three variables, is first solved to obtain the angular velocity ω for the first system of equations from the solution of the variable x4. The latter is then entered as a factor alongside the other factors into the first system of equations, so that the first system of equations comprises two equations with two variables. Alternatively, the angular velocity ω for the first system of equations can be obtained based on a measuring device for determining the angular position or the rotational speed of the rotor, by including a time derivative of the determined angular position.

[0017] According to another embodiment, the state representation comprises a combination of the first system of equations and the second system of equations, resulting in a third system of equations with the following equations: x1,k=x1,k−1⋅(1−Rs⋅TsLd)+x2,k−1⋅Lq⋅Ts⋅x4,k−1Ld+u1⋅TsLd x2,k=x1,k−1⋅(1−Ld⋅Ts⋅x4,k−1Lq)+x2,k−1⋅(1−Rs⋅TsLq)+u2⋅TsLq−k⋅Ts⋅x4,k−1Lq x3,k=x3,k−1+x4,k−1⋅Ts+x5,k−1⋅Ts22 x4,k=x4,k−1+x5,k−1⋅Ts x5,k=x5,k−1 In contrast to the state representation described above, the third system of equations contains five equations with five variables, all of which can be solved simultaneously.

[0018] The state representation can be pre-stored in an external storage medium to which the device or control unit executing the control method according to the invention has access, or in the control unit's own storage medium. The device or control unit can be an integral component of a central control unit of the vehicle, such as the electronic control unit (ECU), or alternatively communicate with the central control unit or the ECU.

[0019] Alternatively or additionally, the state representation or mathematical representation (e.g., the state-space representation) can be obtained from a partially or fully semantic description of the operating state of the electric drive unit. In this case, the device or control unit according to the invention can be configured to extract the rotor state variable(s) and / or the stator state variable(s) from the partially or fully semantic description using a suitably adapted and / or trained algorithmic model, such as a corresponding neural network, for example, a deep learning model or a large-language model (LLM). In the case of the LLM, the latter comprises several layers of neurons, which are structured according to a transformer architecture with encoders and decoders, preferably additionally with a self-attention mechanism.The algorithmic model has preferably been trained to extract a corresponding mathematical description of the stator and / or rotor state variable(s) as a state representation from the partially or fully semantic description.

[0020] The method according to the invention comprises a further process step of generating an estimated result of the state parameter or several state parameters from the received state representation by means of an observer module. The observer module can be a computing module integrated into the device executing the method according to the invention or into the control unit, or it can be a separate computing module connected to the device or control unit via a data link. The computing module uses an algorithmic model, which is to be understood as an observer model. The state representation is input into the observer module, which is preferably configured to generate the estimated result by solving the first system of equations and / or the second system of equations and / or the third system of equations.The number of calculation steps k initially assumes a starting value of 1, which in the case of the third system of equations (which applies analogously to the case of the first or second system of equations) corresponds to the initial variables x. 1,0 , x 2,0 , x 3,0 , x 4,0 , x 5,0 This corresponds to variables whose values ​​are preferably predefined. In the next calculation step, the variables x 1,1 , x 2,1 , x 3,1 , x 4,1 , x 5,1 based on the initial variables x 1,0 , x 2,0 , x 3,0 , x 4,0 , x 5,0 by solving the system of equations used. In this way, the variables x 1,k , x 2,k , x 3,k , x 4,k , x 5,k of the current (or k-th) calculation step based on the variables x 1,k-1 , x 2,k-1 , x 3,k-1 , x 4,k-1 , x 5,k-1The result of the previous (or (k-1)th) calculation step is determined by solving the system of equations used. This step-by-step solution of the system of equations is equally applicable to the first and the second system of equations.

[0021] The method according to the invention comprises a further process step of receiving a detection result relating to the state parameter or several state parameters. The detection result is obtained by means of one or more sensor devices. In particular, the detection result can comprise direct sensor data. For example, the detection result for the rotor state variables angular position and rotational speed can be obtained by means of rotor state sensors, such as Hall sensors. In addition, the detection result for the stator state variables, in particular the phase currents, can be obtained by current sensors. Alternatively or additionally, the detection result can be obtained by further processing the direct sensor data. For example, the detection result for the d- and / or q-current component can be obtained by transforming phase current data. In addition, the detection result for the angular velocity or...The angular acceleration can be obtained by taking the time derivative(s) of the angular position data.

[0022] The acquisition result can comprise a sensor data record in the case of a single sensor device, or, in the case of multiple sensor devices, multiple sensor data records, one for each of the multiple sensor devices. Within a single, preferably each, data record, multiple sub-data records can exist that refer to the same state parameter and follow each other chronologically. Alternatively or additionally, the sub-data records can each refer to their own separate state parameter.

[0023] The data acquisition result exhibits an error pattern. This error pattern preferably contains at least one of the following error pieces of information: - an initial error message about a faulty sensor data set that is assigned to one of the sensor devices; - a second error message about at least one of the sensor devices that is faulty, for example by having a defect and / or a partial or complete failure; - a third error information about a quantitative property of the acquisition result, wherein the quantitative property includes one or more evaluation parameters that are suitable to quantitatively describe a degree of the error pattern, wherein the evaluation parameters include, for example, a noise component, an offset, a phase shift and / or a degree of asymmetry; wherein, in the event that the faulty sensor data set contains several sub-data sets, the fault information preferably further includes a fourth fault information about one of the sub-data sets that is faulty.

[0024] The fault pattern can be determined by detecting / evaluating the acquisition result (preferably continuously and / or in real time) using the device or control unit according to the invention. Alternatively, the fault pattern can be provided by an external unit, such as a device monitoring the sensor devices (preferably continuously and / or in real time). The degree of severity can be determined based on the evaluated parameter(s). In the case of a single evaluated parameter, the degree of severity can be determined directly from the value of the evaluated parameter. In the case of multiple evaluated parameters, the degree of severity can be determined by averaging the values ​​of the evaluated parameters, in particular by weighted averaging, in which the respective evaluated parameters are weighted (i.e.,The values ​​of the respective evaluation parameters are determined (multiplied by an associated weighting factor). However, the present invention is not limited to this.

[0025] The method according to the invention comprises a further process step of adapting the observer module based on the obtained error pattern. The adaptation of the observer module is preferably based on the quantitative property, in particular on the value of the considered evaluation parameter(s) (e.g., the detected noise component) or the severity of the error pattern. For this purpose, the determined quantitative property, in particular the value of the considered evaluation parameter(s) or the severity of the error pattern, preferably the detected noise component, is compared, for example, with a predefined threshold value. If the quantitative property comprises several evaluation parameters, the value of each evaluation parameter is compared with a corresponding predefined threshold value. Depending on the comparison result, the observer module is adapted or adjusted.

[0026] According to one embodiment, the observer module is adapted such that a proportion of the faulty sensor data set and / or the at least one faulty sub-data set and / or the sensor data set of the faulty sensor device ultimately depends on the quantitative property, in particular on the value(s) of the evaluation variable(s) or the degree of the fault pattern, preferably on the noise component. In the case of the "AND" operation, a separate proportion is assigned to each faulty sensor data set, each faulty sub-data set, and each sensor data set of the faulty sensor device.The proportion factor indicates the degree to which the faulty sensor data set and / or the faulty sub-data set and / or the sensor data set of the faulty sensor device is to be considered in a final result of the state parameter or multiple state parameters, to be determined later using the adapted observer module. Depending on the severity of the fault pattern, in particular the high / low value of the respective evaluation variables (e.g., the noise component) or the degree of the fault pattern, the respective sensor data set is to be weighted and considered accordingly. Preferably, the proportion factor decreases with the detected noise component. The more pronounced the fault pattern, e.g., the higher the value of the respective evaluation variables under consideration (e.g., the noise component) or the degree of the fault pattern, the more "unreliable" or "unusable" it is for generating the control signal.The control signals are linked to their corresponding sensor data sets, which are therefore weighted with a correspondingly lower factor in the final result. The proportion factor can be a weighting factor for the respective sensor data set, which is then used to consider the state parameter(s) in the final result, as described in more detail below. Alternatively or additionally, the proportion factor is based on one or more covariance matrices of a Kalman filter, which is preferably implemented algorithmically in the observer module.

[0027] According to a further embodiment, if the value of the respective evaluation parameters under consideration, e.g., the noise component, or the severity of the error pattern exceeds the predefined threshold, the observer module is adjusted to reduce the proportion factor in the final result, in particular to zero. Alternatively or additionally, if the value of the respective evaluation parameters under consideration, e.g., the noise component, or the severity of the error pattern falls below the predefined threshold, the observer module is adjusted to base the final result of the state parameter(s) on an error-free portion of the acquisition result. In this way, the "usable" information content from the acquisition result is used to generate the control signals.

[0028] According to a further embodiment, if the received detection result does not show the fault pattern or only shows it partially, the adapted observer module is accordingly fully or partially reverted to the original observer module before the adaptation. This takes into account the situation where the (preferably continuously monitored / detected) fault pattern is no longer present, and the detection result must be fully considered in the final result of the state parameter(s).

[0029] The method according to the invention comprises a further process step of determining a final result for the state parameter(s) based on the estimation result and the acquisition result using the adapted observer module. The (adapted) observer module is preferably configured to relate the estimation result and the acquisition result to each other in order to determine the final result using a recursive algorithm based on the estimation result and the acquisition result. The observer module includes, for example, a Kalman filter having a predictor-corrector structure. Using the Kalman filter, a value and / or a range of values ​​for the respective state parameter to be determined can be identified, corresponding to a predefined confidence interval.The identified value / range of values ​​is therefore probability-optimized, so that the state parameter lies within the identified value or range of values ​​with a reliable probability. Because the number of data inputs required for the Kaiman filter is comparatively small, the state parameters can be determined in this way with significantly reduced computational effort and increased reliability.

[0030] The calculated final result is output and used to generate one or more control signals for controlling the power output stage based on the estimation result. In contrast to methods known from the prior art, the input data required to provide the control signal for switching the power output stage or the inverter is reduced. In this way, the control signal can be generated particularly easily according to the invention. Furthermore, complex replacement measures are unnecessary to resume operation without disruption, thus enabling a smooth transition from the failure state to normal operation of the drive system without abrupt fluctuations in the state parameters or shutdown of the electric drive unit.

[0031] The invention further relates to the electric axle drive for the at least partially electrified vehicle with the device (or control unit) according to the invention, the at least partially electrified vehicle with the electric axle drive according to the invention, and a computer-readable (storage) medium comprising commands which, when executed by a computer, cause the computer to perform the steps of the method according to the invention in any one of the embodiments described herein. The advantages already described in connection with the method according to the invention also apply to the device, the electric axle drive, the vehicle, and the storage medium according to the invention.

[0032] The features of the claims and the features described above and below with reference to the drawings complement each other. Features that become apparent in the exemplary embodiments, both individually and in each combination of features, advantageously further develop the subject matter of the claims and also the embodiments described above.

[0033] The invention is explained below by way of example with reference to embodiments shown in the figures.

[0034] They show: Fig. 1 a schematic representation of an electric axle drive in a vehicle that is at least partially electrified; Fig. 2 a schematic representation of a method for controlling the electric axle drive for the at least partially electrified vehicle; Fig. 3 a schematic circuit diagram of a method for determining rotor state variables and / or stator state variables of the electric axle drive using an observer module.

[0035] Identical objects, functional units, and comparable components are designated across all figures using the same reference symbols. These objects, functional units, and comparable components are identical in their technical characteristics unless explicitly or implicitly stated otherwise in the description.

[0036] Fig. Figure 1 shows, purely schematically and by way of example, a partially electrified vehicle 50, which includes an electric axle drive 40. The electric axle drive 40 has an electric drive unit 42 and a power output stage 46. The power output stage 46 is preferably designed as a DC / AC inverter and has several half-bridges corresponding to the phases of a stator of the electric drive unit 42 (not shown in detail here). In each half-bridge, a high-side device consisting of one or more parallel-connected circuit breakers and a low-side device consisting of one or more parallel-connected circuit breakers are connected in series between different potentials of an intermediate circuit voltage. The corresponding phase of the electric drive unit 42 is contacted where the high-side device and the low-side device are connected.If the high-side and low-side circuit breakers are alternately opened and closed with a predetermined duty cycle, the desired phase voltage is established at the respective phase, which is then supplied to the corresponding coil winding. In this way, pulse-width modulation (PWM) can be used to generate phase voltages that vary sinusoidally over time at the individual phases of the electrical drive unit 42. The electrical drive unit 42 is therefore energized and driven based on the control of the power output stage 46 or the inverter. Preferably, this is achieved using field-oriented control (FOR), which enables the decoupling of the field- and torque-generating components, also referred to as the d- and q-components, respectively.In this process, stator state variables (especially phase currents of the stator coil windings) and rotor state variables (especially angular position, velocity, acceleration, and / or rotor speed) are acquired and transformed into a rotating coordinate system (dq coordinate system). Data for the d-current component and the q-current component are then generated from the transformed phase current data.

[0037] Fig. Figure 2 shows a schematic representation of an exemplary method 100 for controlling the electric axle drive 40. The method 100 comprises several process steps 101-106, which are carried out by means of the device 44 according to the invention or the control unit 44 of the electric axle drive 40. The device 44 or the control unit 44 can be an integral component of a central control unit of the vehicle, such as the electronic control unit (ECU), or alternatively communicate with the central control unit or the ECU.

[0038] The procedure 100 comprises a procedure step 101 in which a state representation 12 (see Fig. 3), which describes an operating state of the electric axle drive 40, in particular the electric drive unit 42, is received. The state representation 12 relates to one or more state parameters of the electric drive unit 42. The method 100 comprises a further method step 102 in which an estimation result 14 (see Fig. 3) of the state parameter or the multiple state parameters from the state representation 12 using an observer module 30 (see Fig. 3) is generated. In a further process step 103, a detection result 16 concerning the state parameter or the multiple state parameters is received. In a further process step 104, an error pattern 18 of the detection result 16 is obtained. The process 100 comprises a further process step 105 of adapting the observer module 30 based on the error pattern 18. Finally, in a further process step 106, a final result 34 of the state parameter or the multiple state parameters is determined based on the estimation result 14 and the detection result 16 using the adapted observer module 30.

[0039] The state representation 12 preferably describes, by means of a mathematical representation (e.g., a state-space representation), one or more state parameters of the electric drive unit 42, which preferably include one or more rotor state variables such as an angular position, an angular velocity, an angular acceleration, and a rotor speed, and one or more stator state variables such as one or more phase currents, a d-current component, and / or a q-current component. Depending on the state parameters, the state representation 12 describes the operating state of the electric drive unit 42.

[0040] The state representation 12 preferably comprises one or more systems of equations, more preferably differential systems of equations. The state representation 12 can be pre-stored in an external storage medium to which the device 44 executing the control method according to the invention or the control unit 44 has access, or in its own storage medium of the control unit 44. The device or the control unit 44 can be an integral component of a central control unit of the vehicle 50, such as the electronic control unit (ECU), or alternatively communicate with the central control unit or the ECU.

[0041] The measurement result 16 is obtained using one or more sensor devices. The measurement result 16 can include direct sensor data. For example, the measurement result 16 for the rotor state variables angular position and rotational speed can be obtained using rotor state sensors, such as Hall sensors. Furthermore, the measurement result 16 for the stator state variables, in particular the phase currents, can be obtained using current sensors. Alternatively or additionally, the measurement result 16 can be obtained by further processing the direct sensor data. For example, the measurement result 16 for the d- and / or q-current component can be obtained by transforming phase current data. Furthermore, the measurement result 16 for the angular velocity or angular acceleration can be obtained by taking the time derivative(s) of the angular position data.

[0042] The acquisition result 16 can comprise a sensor data record in the case of a single sensor device, or, in the case of multiple sensor devices, multiple sensor data records, one for each of the multiple sensor devices. Within one, preferably each, data record, multiple sub-data records can exist that refer to the same state parameter and follow each other chronologically. Alternatively or additionally, the sub-data records can each refer to their own state parameter.

[0043] How Fig.As shown in Figure 3, purely by way of example and schematically, both the received state representation 12 and the received acquisition result 16 are entered into the observer module 30. The observer module 30 can be a computing module integrated into the device 100 executing the method according to the invention or into the control unit 44, or it can be a separate computing module connected to the device or control unit 44. The computing module uses an algorithmic model, which can be understood as an observer model. The observer module 30 is preferably configured to generate the estimation result 14 by solving the system or systems of equations.

[0044] Error pattern 18 preferably contains at least one of the following error information: - an initial error message about a faulty sensor data set belonging to one of the sensor devices; - a second error message about at least one of the sensor devices that is faulty; - a third error information about a quantitative property of the acquisition result, which includes one or more evaluation parameters (e.g., a noise component, an offset, a phase shift and / or an asymmetry) that are suitable for quantitatively describing a degree of severity of the error pattern; wherein, in the case that the faulty sensor data set contains several sub-data sets, the error information preferably also includes a fourth error information about one of the sub-data sets that is faulty.

[0045] The fault pattern 18 can be detected by the device or control unit 44 according to the invention by means of (preferably continuous and / or real-time) detection / evaluation of the acquisition result 16. Alternatively, the fault pattern 18 can be provided by an external unit, such as a device that monitors the sensor devices (preferably continuously and / or in real time).

[0046] The adjustment of the observer module 30 based on the error pattern 18 is preferably carried out based on the previously determined / detected noise component contained in the detection result 16 within the error pattern 18. For this purpose, the value of the evaluated parameter under consideration, such as the detected noise component, or the severity of the error pattern is compared with a predefined threshold value. Depending on the comparison result, the observer module 30 is adjusted or calibrated.The observer module 30 can preferably be adapted such that a proportion factor of the faulty sensor data set and / or of the at least one faulty sub-data set and / or of the sensor data set of the faulty sensor device in the final result 34 depends on the quantitative property, in particular on the value of the respective evaluation variables considered, such as the noise component, or on the severity of the fault pattern, and further preferably decreases with the quantitative property, in particular with the value of the respective evaluation variables considered, such as the noise component, or with the severity of the fault pattern. In the case of the "AND" operation, a separate proportion factor is assigned to each faulty sensor data set, the faulty sub-data set, and the sensor data set of the faulty sensor device.The proportion factor indicates the degree to which the faulty sensor data set and / or the faulty sub-data set and / or the sensor data set of the faulty sensor device is to be considered in a final result 34 of the state parameter or several state parameters, to be determined later by means of the adapted observer module 30. Depending on the severity of the fault pattern, in particular how high / low the value of the respective evaluation variables (e.g., the noise component) is, the respective sensor data set is to be weighted and considered accordingly. Preferably, the proportion factor decreases with the severity of the fault pattern, in particular with the value of the evaluation variable(s) considered, e.g., the detected noise component. The more pronounced the fault pattern, in particular the higher the value of the evaluation variable(s) considered, e.g., the noise component, the more "unreliable" or "less reliable" the result is.The associated sensor data set is "less suitable" for generating the control signal(s) 44, so it must be considered in the final result with a correspondingly lower weighting factor. The proportion factor can be a weighting factor for the respective sensor data set, with which the latter is considered in the final result 34 for the state parameter(s), as described in more detail below. Alternatively or additionally, the proportion factor is based on one or more covariance matrices of a Kalman filter, which is preferably implemented algorithmically in the observer module 30.

[0047] The (adapted) observer module 30 is preferably configured to relate the estimation result 14 and the acquisition result 16 to each other in order to determine the final result 34 using a recursive algorithm based on the estimation result 14 and the acquisition result 16. The observer module 30 includes, for example, the aforementioned Kalman filter (or is configured as a Kalman filter), which has a predictor-corrector structure 32. Using the Kalman filter, a value and / or a range of values ​​for the respective state parameter to be determined can be identified, corresponding to a predefined confidence interval. The value / range of values ​​thus identified is therefore probability-optimized, such that the state parameter lies within the identified value or range of values ​​with a reliable probability.Because the number of data inputs required for the Kalman filter is relatively small, the state parameters can be determined in this way with significantly reduced computational effort and increased reliability.

[0048] The determined final result 34 is output and used to generate one or more control signals for controlling the power output stage 46 based on the estimation result. In contrast to methods known from the prior art, complex compensatory measures to overcome the defect / failure of one or more sensor devices are unnecessary. Driving operation can be resumed without disruption, enabling a smooth transition from the failure state to normal operation of the drive system without abrupt fluctuations in the state parameters or shutdown of the electric drive unit 42. Reference sign 12 State Representation 14 Estimate result 16 Recording result 18 Error pattern 30 observer modules 32 Predictor-Corrector Structure 34 Final result 40 electric axle drive 42 electric drive unit (electric machine) 44 Device (control unit) 46 Power output stage (inverter) 50 at least partially electrified vehicles 100 procedures 101-106 Procedural steps

Claims

[1] Method (100) for controlling an electric axle drive (40) in a vehicle (50) that is at least partially electrified, wherein the electric axle drive (40) comprises an electric drive unit (42) and a power output stage (46), the method (100) comprising: - Receiving a state representation (12) concerning one or more state parameters of the electric drive unit (42); - Generating an estimation result (14) of the state parameter or several state parameters from the state representation (12) using an observer module (30); - Receiving a detection result (16) concerning the state parameter or the several state parameters, wherein the detection result (16) is obtained by means of one or more sensor devices; - Obtaining an error image (18) of the recording result (16); - Adjusting the observer module (30) based on the error pattern (18); - Determining a final result (34) of the state parameter or several state parameters based on the estimation result (14) and the acquisition result (16) using the adapted observer module (30), wherein the final result (34) is intended to provide a control signal for controlling the power output stage (46). [2] Method (100) according to claim 1, wherein the error pattern (18) contains at least one of the following error information: - an initial error message about a faulty sensor data set belonging to one of the sensor devices; - a second error message about at least one of the sensor devices that is faulty; - a third error information about a quantitative property of the acquisition result, comprising one or more evaluation parameters suitable for quantitatively describing the degree of the error pattern; wherein, in the case that the faulty sensor data set contains several sub-data sets that follow one another in time and / or that relate to different state parameters, the error information preferably further comprises a fourth error information about one of the sub-data sets that is faulty. [3] Method (100) according to claim 2, wherein the adjustment of the observer module (30) is based on the quantitative property, preferably wherein the value of the considered evaluation variable(s) or the degree of the error pattern is compared with a predefined threshold value. [4] Method (100) according to claim 3, wherein the observer module (30) is adapted such that a proportion factor of the faulty sensor data set and / or of the at least one faulty sub-data set and / or of the sensor data set of the faulty sensor device in the final result depends on the quantitative property, in particular on the value of the evaluated variable(s) considered or on the degree of the fault pattern, e.g. on the noise component, preferably decreases with the degree of the fault pattern. [5] Method (100) according to claim 4, wherein, in the event that the value of the evaluation variable(s) or the degree of the error pattern exceeds the predefined threshold, the observer module (30) is adjusted to reduce the proportion factor in the final result (34), in particular to zero. [6] Method (100) according to one of claims 3 to 5, wherein, in the event that the value of the evaluation variable(s) or the degree of the error pattern falls below the predefined threshold, the observer module (30) is adjusted to base the final result (34) on an error-free part of the recording result (16). [7] Method (100) according to one of the preceding claims, wherein, in the event that the received detection result (16) does not exhibit the error pattern or only exhibits it partially, the adapted observer module (30) is accordingly fully or partially reverted to the original observer module (30) before the adaptation. [8] Method (100) according to one of the preceding claims, wherein the observer module (30) is configured to relate the estimation result (14) and the recording result (16) to each other in order to determine the final result (34). [9] Method (100) according to any of the preceding claims, wherein the observer module (30) comprises a Kalman filter (32). [10] Device (44), in particular control unit, for controlling an electric axle drive (40) in an at least partially electrified vehicle (50), wherein the device (44) is configured to perform the method (100) according to any one of claims 1 to 9. [11] Electric axle drive (40) for a vehicle (50) that is at least partially electrified, comprising: - an electric drive unit (42); - a power output stage (46), in particular an inverter; and - a device (44) according to claim 10. [12] At least partially electrified vehicle (50) with an electric axle drive (40) according to claim 11. [13] Computer-readable storage medium comprising instructions which, when executed by a computer, cause it to perform the steps of the method (100) according to any one of claims 1 to 9.

Images