Method and device for operating a vehicle dynamics system

Abstract

The invention relates to a method for a driving dynamics system (1) operated in a motor vehicle, wherein a driving dynamics function is executed; the method comprises the following steps: - determining a modeled control intervention (S) on the basis of a data-based trainable driving dynamics model (21) and as a function of input values ​​(E) describing the driving state of the motor vehicle; - limiting one or more output values ​​of the modeled control intervention (M) by means of a limiting function (22) in order to obtain a monitored control intervention for intervening in a braking system and / or a drive system of the motor vehicle, wherein the limiting function (22) corresponds to a parameterized rule-based driving dynamics function which defines a limiting range (B) for each of the one or more output values ​​as a function of the input value (E); - operating the driving dynamics system with the aid of the monitored control intervention (S).

Description

Technical Field

[0001] The present invention relates to a vehicle dynamics system and in particular to a regulating device for a vehicle dynamics system. Background Art

[0002] Driving dynamics systems, such as ABS (anti-lock braking system), TCS (traction control system) and ESP (electronic stability program), intervene in the driving functions of the motor vehicle by means of control interventions in order to improve driving stability in different driving situations. Such control interventions usually include braking interventions at one or more wheels and / or adaptation of the setpoint engine torque to be set.

[0003] Driving dynamics systems usually have functions that are defined by a very large number of applicable parameters. These parameters are usually determined manually during test drives and applied as a function of the observed driving behavior. This makes the application process for driving dynamics systems very complex and must be carried out again individually for each vehicle model.

[0004] Until now, there have been no plans to use machine learning methods, such as neural networks, for operating a driving dynamics system, since the driving dynamics function or driving dynamics function is safety-critical due to its control interventions on the brake system and drive system, and machine learning functions do not meet the requirements for the testability of safety-critical software.

[0005] However, the use of a data-based, trainable model can significantly reduce the parameterization effort and, moreover, can achieve improved properties of the vehicle dynamics system. Summary of the invention

[0006] According to the invention, a method for operating a vehicle dynamics system as well as a vehicle dynamics system and a device are provided.

[0007] Other technical solutions include:

[0008] 1) The input value includes one or more of the following values:

[0009] - the rotational speed of one or more of the wheels;

[0010] - the brake pressure at the brake pedal or in the brake circuit of the brake device or at the wheels;

[0011] - acceleration in the longitudinal direction of the vehicle and in the transverse direction of the vehicle;

[0012] - the yaw rate during the turning turn of the steerable wheels;

[0013] - Steering angle;

[0014] - accelerator pedal position of the accelerator pedal;

[0015] - a measured or modeled motor drive torque of the drive motor; and

[0016] - the measured motor speed of the drive motor.

[0017] 2) The one or more output values ​​of the regulatory intervention include one or more of the following values:

[0018] - a braking pressure or force for one or more of said wheels; and

[0019] A setpoint drive torque for controlling the drive motor.

[0020] 3) The data-based trainable driving dynamics model comprises a Gaussian process model and / or a neural network.

[0021] 4) The limiting function has a parameterized rule-based control function in order to model one or more output values ​​of the control intervention, wherein a limiting range is determined for each of the one or more output values ​​by specifying a tolerance range.

[0022] 5) The neural network has a regression output layer.

[0023] 6) The neural network is a deep neural network.

[0024] According to a first aspect, a method for operating a vehicle dynamics system in a motor vehicle is provided, wherein a vehicle dynamics function or a vehicle dynamics function is executed; the method comprises the following steps:

[0025] - determining a modeled control intervention on the basis of a data-based trainable driving dynamics model and as a function of input values ​​indicating the driving state of the motor vehicle;

[0026] - limiting one or more output values ​​of the modeled control intervention by means of a limiting function in order to obtain a monitored control intervention for intervening in a braking system and / or a drive system of the motor vehicle, wherein the limiting function corresponds to a parameterized rule-based driving dynamics function which defines a limiting range for each of the one or more output values; and

[0027] The vehicle dynamics system is operated with the aid of the monitored control interventions.

[0028] One concept of the method for operating a vehicle dynamics system is to implement a vehicle dynamics function based on a data-based vehicle dynamics model. The data-based vehicle dynamics model can be trained with the aid of data from real driving maneuvers and usage measurements and on the basis of suitable control interventions, the data from driving maneuvers and usage measurements representing input values ​​for the vehicle dynamics model, so that the vehicle dynamics model can provide respective modeled control interventions for different driving situations.

[0029] The modeled control interventions of the data-based driving dynamics model are limited by means of a limiting function or limiting function, which is essentially based on a classical driving dynamics control. The limiting function provides, on the basis of the input values, upper and lower limits for the modeled control interventions of the driving dynamics model, to which the modeled control interventions are limited. The limiting function can be similar to a conventional driving dynamics control, i.e. with a parameterized, rule-based driving dynamics model, but it can be parameterized with significantly fewer parameters, since only safety-threatening values ​​of the modeled control values ​​should be suppressed by the limiting. The otherwise common fine-tuning of the conventional driving dynamics model is not necessary for implementing the limiting function.

[0030] The method described above can significantly simplify the effort for implementing the driving dynamics control and, in particular, significantly reduce the effort for parameterization. The data-based driving dynamics model can be used to adapt the driving dynamics function or driving dynamics function to an individual vehicle or vehicle type by evaluating the actual driving situation. Since the limiting function is usually defined heuristically or based on a model / rule, in particular in the form of a classical controller, the driving dynamics function or driving dynamics function implemented by the method described above can be checked in a classical manner.

[0031] In particular, the input value may include one or more of the following values:

[0032] - the rotational speed of one or more of the wheels;

[0033] - the brake pressure at the brake pedal or in the brake circuit or at the wheels;

[0034] - acceleration in the longitudinal direction of the vehicle and in the transverse direction of the vehicle;

[0035] - the yaw rate during the turning turn of the steerable wheels;

[0036] - Steering angle;

[0037] - accelerator pedal position of the accelerator pedal;

[0038] - measured or modeled motor drive torque; and

[0039] - the measured motor speed of the drive motor.

[0040] Furthermore, the one or more output values ​​of the control intervention may include one or more of the following values:

[0041] - a braking pressure or force for one or more of said wheels; and

[0042] A setpoint drive torque for controlling the drive motor.

[0043] It may be provided that the data-based, trainable driving dynamics model comprises a neural network, in particular a deep neural network, in particular with a regression output layer.

[0044] According to one specific embodiment, the limiting function or limiting function may have a parameterized rule-based regulating function or regulating function in order to model one or more output values ​​of the regulating intervention, wherein a limiting range is determined for each of the one or more output values ​​by specifying a tolerance range.

[0045] According to a further aspect, a device for operating a driving dynamics system in a motor vehicle, in particular a control unit, is provided, in which a driving dynamics function or a driving dynamics function is executed; wherein the device is designed to: - determine a modeled control intervention on the basis of a data-based trainable driving dynamics model and as a function of input values ​​indicating a driving state of the motor vehicle;

[0046] - limiting one or more output values ​​of the modeled control intervention by means of a limiting function in order to obtain a monitored control intervention for intervening in a brake system and / or a drive system of the motor vehicle, wherein the limiting function corresponds to a parameterized rule-based driving dynamics function which defines a limiting range for each of the one or more output values;

[0047] The vehicle dynamics system is operated with the aid of the monitored control interventions.

[0048] According to a further aspect, a driving dynamics system is provided for a motor vehicle, which has the above-described device and has a drive motor and / or one or more brake devices at the wheels of the motor vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] The following is a further explanation of the embodiment with the aid of the accompanying drawings. The accompanying drawings show:

[0050] Figure 1 A schematic illustration of a driving dynamics system;

[0051] Figure 2 a block diagram for illustrating a control system implemented in a control unit of the vehicle dynamics system; and

[0052] Figure 3 In the example of ABS modulation, the actually set brake pressure and the profile of the data-based driving dynamics function and of the limiting function. DETAILED DESCRIPTION

[0053] Figure 1 A schematic illustration of a driving dynamics system 1 for a motor vehicle is shown. Figure 1 The drive system and brake system for the four wheels of a motor vehicle are shown, which together with the corresponding control units form the vehicle dynamics system. For this purpose, a drive motor 2 is connected to the wheels 4 via an output line 3 in order to provide a drive torque for driving the wheels 4. In addition, the wheels 4 are equipped with brake devices 5, in particular friction brakes, which can be individually actuated.

[0054] The drive motor 2 and the brake device 5 can be controlled by a control unit 6. The control unit 6 is also used to implement driving dynamics functions, in particular ABS (anti-lock braking system), TCS (traction control system) and ESP (electronic stability program). The control unit 6 can be implemented in a motor controller and combined with another motor control device or a control device of another equipment. Alternatively, the control unit 6 can also be designed separately. The mode of operation of the driving dynamics function is known and is usually implemented using a parameterizable driving dynamics function, which is based on a conventional control function or control function or includes such a conventional control function or control function.

[0055] Input values ​​for executing the driving dynamics function are supplied to the control unit 6 via an internal system model, internal sensors or an external sensor system 7. The input values ​​may include, for example, one or more of the following values:

[0056] - the rotational speed of one or more wheels 4;

[0057] - the brake pressure at the brake pedal 8 or in the brake circuit or at the wheel 4;

[0058] - acceleration in the longitudinal direction of the vehicle and in the transverse direction of the vehicle, in particular detected by inertial sensors;

[0059] - the yaw rate during the steering turn of the steerable wheel 4;

[0060] - Steering angle;

[0061] - accelerator pedal position of accelerator pedal 9;

[0062] - measured or modeled motor drive torque; and

[0063] The measured motor speed of drive motor 2 .

[0064] The driving dynamics function provides at least one of the following output values ​​as one or more output values ​​of the control intervention:

[0065] - a braking pressure or a braking force for one or more of the wheels 4; and

[0066] A setpoint drive torque for controlling the drive motor 2 .

[0067] Figure 2 A block diagram is shown for explaining the driving dynamics function executed in the control unit 6. The driving dynamics function is formed with the aid of a trained, data-based driving dynamics model 21 and with a limitation function 22, which performs parallel calculations in order to determine a modeled control intervention.

[0068] The data-based driving dynamics model 21 comprises a trainable model (in particular a neural network), a regression model (for example a Gaussian process model), etc. The trainable model is trained in a suitable manner in order to obtain one or more output values ​​of the control intervention M based on one or more of the input values ​​E as output values ​​of the control intervention.

[0069] The input value or values ​​E are simultaneously supplied to the limiting function or limiting function 22 , which contains a conventional embodiment of the vehicle dynamics control.

[0070] The limiting function represents a simply parameterized, rule-based driving dynamics function which provides upper and lower limit values. The parameterized, rule-based driving dynamics function can include, for example, a PID control function or the like. The limiting function provides lower and upper limit values ​​for the modeled control intervention, wherein the implementation of the limiting function or limiting function is simpler than the implementation of a conventional driving dynamics function or driving dynamics function, since, due to the output of the limit range (value channel), the output value of the control intervention does not need to be determined in an accurately fine-tuned manner.

[0071] The lower limit value and the upper limit value of the limiting function 22 represent a limiting range, which ensures a value path for acceptable performance of the driving dynamics function. Alternatively, this can be achieved by acquiring one or more approximate output values ​​of the output value as a control intervention, which are acquired by a classical system with fewer parameters and in each case a tolerance range is determined, the upper and lower limits of which indicate the upper and lower limit values ​​of the limiting range, respectively, within which the optimum value should lie.

[0072] The modeled control intervention M of the data-based driving dynamics function is limited to a limit range B specified by the limit function for each of its control intervention output values ​​by means of the coordinator 23. The limited values ​​of the output values ​​of the control intervention S are then used as the control intervention for the driving dynamics function, and the brake system and / or the drive system are controlled accordingly.

[0073] exist Figure 3 , the variation curve of the brake pressure P as an exemplary output value of the control intervention S over time is schematically shown. The variation curves of the upper limit value BO and the lower limit value BU for the brake pressure P provided by the limiting function 22 can be seen. In addition, the variation curve of the modeled control intervention M of the data-based driving dynamics model is represented by a dotted line. With the help of the dotted line, it can be seen that the output value of the brake pressure provided by the data-based driving dynamics model is used as long as the corresponding value exists within the limiting range B. If the value of the brake pressure provided by the data-based driving dynamics model exists outside the limiting range B, then this value is limited to the upper limit value or the lower limit value B. The limited control intervention S is represented in the form of a dotted line.

[0074] In this way, the use of conventional vehicle dynamics control can be simplified, since the number of parameters to be defined is reduced. The coordinator ensures that the value for the control intervention output value is never set outside a limit range B that is already determined by the limit function 22 based on the conventional vehicle dynamics control. The function of the coordinator 23 is rule-based and can therefore be verified by classical methods, so that despite the use of a machine learning model as a data-based vehicle dynamics model, detectability is not critical.

[0075] For the data-based driving dynamics model, a fully connected feedforward neural network (multi-layer perceptron (MLP)) with a regression output layer is preferably provided. The training data for training the driving dynamics model can be extracted from a database with driving maneuvers and usage measurements, which come from the real operation of the motor vehicle. In addition, the above-mentioned input values ​​are mapped to the output value or values. In addition, vehicle parameters that indicate the type of the relevant vehicle can also be supplied to the data-based driving dynamics model.

Claims

1. A method for operating a driving dynamics system (1) in a motor vehicle, wherein: A driving dynamics function is implemented; it has the following steps: - determining a modeled control intervention (S) on the basis of a data-based trainable driving dynamics model (21) and as a function of input values ​​(E) describing the driving state of the motor vehicle; - limiting one or more output values ​​of the modeled control intervention (M) by means of a limiting function (22) in order to obtain a monitored control intervention for intervening in a brake system and / or a drive system of the motor vehicle, wherein the limiting function (22) corresponds to a parameterized, rule-based driving dynamics function which defines a limiting range (B) for each of the one or more output values ​​as a function of the input value (E); The vehicle dynamics system is operated with the aid of the monitored control intervention (S).

2. The method according to claim 1, wherein: The input value (E) includes one or more of the following values: - the rotational speed of one or more of the wheels (4); - a brake pressure at the brake pedal (8) or in the brake circuit of the brake device (5) or at the wheel (4); - acceleration in the longitudinal direction of the vehicle and in the transverse direction of the vehicle; - the yaw rate during the turning turn of the steerable wheels; - Steering angle; - an accelerator pedal position of an accelerator pedal (9); - a measured or modeled motor drive torque of the drive motor (2); and - the measured motor speed of the drive motor (2).

3. The method according to claim 2, wherein: The one or more output values ​​of the regulating intervention (S) include one or more of the following values: - a braking pressure or a braking force for one or more of the wheels (4); and - a rated drive torque for controlling the drive motor (2).

4. The method according to any one of claims 1 to 3, wherein: The data-based trainable driving dynamics model comprises a Gaussian process model and / or a neural network.

5. The method according to any one of claims 1 to 3, wherein: The limiting function (22) has a parameterized rule-based control function for modeling one or more output values ​​of the control intervention, wherein a limiting range (B) is determined for each of the one or more output values ​​by specifying a tolerance range.

6. The method according to claim 4, wherein: The neural network has a regression output layer.

7. The method according to claim 4, wherein: The neural network is a deep neural network.

8. A device for operating a driving dynamics system (1) in a motor vehicle, wherein: A driving dynamics function is implemented; wherein the device is designed to: - determining a modeled control intervention (M) on the basis of a data-based trainable driving dynamics model (21) and as a function of input values ​​(E) which describe the driving state of the motor vehicle; - limiting one or more output values ​​of the modeled control intervention (M) by means of a limiting function (22) in order to obtain a monitored control intervention for intervening in a braking system and / or a drive system of the motor vehicle, wherein the limiting function (22) corresponds to a parameterized, rule-based driving dynamics function which defines a limiting range (B) for each of the one or more output values ​​of the modeled control intervention (M); The vehicle dynamics system is operated with the aid of the monitored control intervention (S).

9. The device for a vehicle dynamics system (1) for operation in a motor vehicle according to claim 8, wherein: The device is a control unit (6).

10. A driving dynamics system (1) for a motor vehicle, comprising a device according to claim 8 or 9 and comprising a drive motor (2) and / or one or more brake devices (5) on wheels (4) of the motor vehicle. 11 . A computer program product comprising a computer program which is configured to execute all the steps of the method according to claim 1 . 12 . An electronic storage medium on which a computer program is stored, which is configured to execute all the steps of the method according to claim 1 .

Images