Method for operating a vehicle and corresponding control unit

Abstract

The invention relates to a method for operating a vehicle and a corresponding control unit, in which, in a prediction phase (102), an estimated pitch dynamics (100a) of the vehicle is predicted using a current longitudinal acceleration (108) of the vehicle and a pitch model (106) of the vehicle, and in an update phase (104), an estimated pitch dynamics (100b) of the vehicle is updated using a pitch model (106) of the vehicle. An updated pitch dynamics (100c) is determined using the divided Kalman filter (112), the estimated pitch dynamics (100a) of the vehicle, and the measured pitch angle rate (110).

Description

Technical Field

[0001] The invention relates to a method for operating a vehicle, a corresponding control unit and a corresponding computer program product. Background Art

[0002] In a vehicle, normal force estimation can be performed based on the vehicle's longitudinal acceleration. For this purpose, the vehicle's mass and center of gravity can be used, for example, to determine the normal forces on the vehicle's axles, particularly during acceleration and braking. Using the normal forces, the transferable acceleration or braking torque for the axles can be determined. Normal force estimation can at least reduce wheel slip or locking.

[0003] EP 0000 03 153 374 B1 describes a normal force estimation. Summary of the Invention

[0004] Against this background, a method for operating a vehicle, a corresponding control unit, and a corresponding computer program product are proposed using the method proposed here. Advantageous developments and refinements of the method proposed here are derived from the description and are described below.

[0005] In the method proposed here, in addition to the measured longitudinal acceleration of the vehicle, the measured pitch rate of the vehicle is also used to determine the normal force on the vehicle axle. The pitch rate is used as a correction variable for the estimated behavior of the vehicle.

[0006] The pitch rate is a very accurate representation of the actual dynamics of the vehicle's motion. For example, the expected pitch rate can be compared with the measured pitch rate to determine the estimated behavior.

[0007] The estimated behavior may have computational errors. For example, the model used for the calculation may have systematic errors. The longitudinal acceleration may also have signal errors. Therefore, there is uncertainty about how realistic the estimated behavior is.

[0008] The pitch rate may also have signal errors such as noise, delay and / or signal drift. Therefore, basing the behavior on direct calculation of the pitch rate may lead to error-prone results.

[0009] In the method proposed here, the estimated behavior and the measured pitch angle rate are balanced against each other using a Kalman filter.

[0010] By means of the method proposed here, errors in the behavior estimation are compensated by the measured pitch angle rate, while signal errors in the pitch angle rate are buffered by the estimated behavior.

[0011] A method for operating a vehicle is proposed, wherein, in a prediction phase, estimated pitch dynamics of the vehicle are predicted using the vehicle's current longitudinal acceleration and a pitch model of the vehicle, and in an update phase, updated pitch dynamics are determined using a partitioned Kalman filter, the estimated pitch dynamics of the vehicle, and a measured pitch angular rate.

[0012] Furthermore, concepts regarding embodiments of the present invention can be considered to be based on the ideas and findings described below.

[0013] A vehicle may have an inertial sensor system. The signals from the inertial sensor system may be input variables for a computational model of the vehicle's behavior. The output variables of the model may, in particular, describe the pitch dynamics of the vehicle. The pitch angle rate may indicate the speed and direction of the pitch. The pitch angle rate is strongly dependent on the actual vehicle properties, such as the vehicle's load and the vehicle's spring characteristics. These actual properties may vary and may only be difficult to describe in a vehicle pitch model.

[0014] The pitch rate can be measured using a 6D sensor system in the vehicle. A 6D sensor system detects linear acceleration in three axes and rotational speed around three axes and maps these into measured values. A 6D sensor system is particularly effective at detecting the actual movement of the vehicle.

[0015] The updated pitch dynamics can be determined using a partitioned Kalman filter. A Kalman filter effectively combines calculated and measured variables. The calculated variables are corrected using the measured variables, with unreliable measured values ​​having little influence on the correction. In the Kalman filter, the expected value of the measured variable is compared with the measured value. The Kalman filter is executed cyclically, performing new calculations with each cycle. These calculations are mutually dependent and tracked using the actual measured variables.

[0016] During the update phase, a residual error can be calculated using the estimated pitch dynamics and the measured pitch angular rate. The estimated pitch dynamics can be corrected using the residual error to obtain updated pitch dynamics. The residual error represents the difference between the expected value and the measured value.

[0017] Using the residuals, the innovation covariance can be calculated. The updated pitch dynamics can be determined using the innovation covariance. The innovation covariance describes the confidence level of the residuals. If the confidence level of the residuals is low, only a small change in the updated pitch dynamics can be made.

[0018] Using the previously determined covariance matrix, the covariance error of the estimated pitch dynamics can be predicted. The estimated covariance matrix can be estimated using the covariance error. Using the estimated covariance matrix and the updated pitch dynamics, an updated covariance matrix can be determined. The covariance matrix can describe the credibility of the basic pitch dynamics. If the previously determined pitch dynamics show little credibility, the estimated pitch dynamics based on this may also have low credibility. The measured pitch angle rate can then have a significant influence on the determination of the updated pitch dynamics and thus improve the credibility of the updated pitch dynamics, which in turn can be described in the updated covariance matrix.

[0019] When the estimated covariance matrix is ​​used, the Kalman gain of the estimated pitch dynamics and the Kalman gain of the measured pitch angular rate can be calculated. The updated pitch dynamics can be determined using the Kalman gain. If the estimated pitch dynamics are credible and the measured pitch angular rate is uncredible, a large Kalman gain of the estimated pitch dynamics and a small Kalman gain of the measured pitch angular rate can be generated. Then, the updated pitch dynamics are more affected by the estimated pitch dynamics than by the measured pitch angular rate. If the estimated pitch dynamics are uncredible and the measured pitch angular rate is credible, a small Kalman gain of the estimated pitch dynamics and a large Kalman gain of the measured pitch angular rate can be generated. Then, the updated pitch dynamics are more affected by the measured pitch angular rate than by the estimated pitch dynamics.

[0020] The Kalman gain can be calculated using an uncertainty matrix of the measurement noise of the measured pitch rate and the process noise of the estimated pitch dynamics. The uncertainty matrix can contain a criterion for the confidence level of the estimated pitch dynamics or the measured pitch rate.

[0021] Using the updated pitch dynamics, the current normal forces on the vehicle's front and rear axles can be determined. Using these normal forces, the acceleration and braking torques on the rear and front axles can be pre-controlled. This prevents vehicle wheel slip and / or locking.

[0022] The method is preferably executed using a computer and can be executed, for example, in the form of software or hardware or in the form of a mixture of software and hardware, for example in an auxiliary system.

[0023] The method proposed here further provides a control unit, wherein the control unit is designed to execute, control or implement the steps of the variant of the method proposed here in a corresponding device.

[0024] A control unit can be an electrical device comprising at least one computing unit for processing signals or data, at least one storage unit for storing signals or data, and at least one interface and / or a communication interface for reading or outputting data embedded in a communication protocol. The computing unit can be, for example, a signal processor, a so-called system ASIC, or a microcontroller, which processes sensor signals and outputs data signals based on the sensor signals. The storage unit can be, for example, a flash memory, an EPROM, or a magnetic storage unit. The storage unit can also be located external to the vehicle, for example in the cloud. The interface can be configured as a sensor interface for reading sensor signals from a sensor and / or as an actuator interface for outputting data signals and / or control signals to an actuator. The communication interface can be configured for wireless and / or wired data reading or output. The interface can also be a software module, such as one located on a microcontroller alongside other software modules.

[0025] Also advantageous is a computer program product or a computer program with a program code, which can be stored on a machine-readable carrier or storage medium, such as a semiconductor memory, hard disk storage, optical storage or cloud storage, and is used to execute, implement and / or control the steps of the method according to one of the above-described embodiments, in particular when the program product or the program is executed on a computer, a control unit or a device.

[0026] It is noted that some of the possible features and advantages of the present invention are described herein with reference to different embodiments. Those skilled in the art recognize that the features of the control unit and method can be combined, adapted or exchanged in an appropriate manner to obtain further embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings, wherein neither the drawings nor the description should be construed as limiting the present invention.

[0028] Figure 1 A diagram showing a method flow according to an embodiment is shown.

[0029] The drawings are schematically only and are not to scale. Identical reference numerals indicate identical or equivalent features. DETAILED DESCRIPTION

[0030] Figure 1 1 shows a diagram of the method flow according to an exemplary embodiment. The method determines the pitching dynamics of the vehicle 100 . The determination is performed cyclically in a prediction phase 102 and an update phase 104 .

[0031] In a prediction phase 102 , an estimated pitch dynamics 100 a of the vehicle is estimated using a pitch model 106 of the vehicle. To this end, the pitch model 106 is supplied with the previous pitch dynamics 100 b and the longitudinal acceleration 108 of the vehicle.

[0032] Updated pitch dynamics 100c of the vehicle are determined in an updating phase 104. For this purpose, estimated pitch dynamics 100a are compared with the vehicle's measured pitch angle rate 110. Pitch angle rate 110 is a supporting variable that is compared with the estimated pitch angle rate of estimated pitch dynamics 100a.

[0033] The updated pitch dynamics 100c will be used as the previous pitch dynamics 100b in the next cycle.

[0034] In the exemplary embodiment, pitch angle rate 110 is measured using a 6D sensor system of the vehicle. The 6D sensor system is designed to detect accelerations on three axes and rotational speeds about three axes and to map these into measured values. Pitch angle rate 110 is the rotational speed about the vehicle's pitch axis or Y axis and describes the speed at which the front of the vehicle rises or falls while the rear of the vehicle is lowering or ascending.

[0035] In an embodiment, the continuous updating of the pitch dynamics 100 is performed using a Kalman filter 112 , in which the estimated pitch dynamics 100 a and the measured pitch angular rate 110 are integrated.

[0036] In an exemplary embodiment, a residual calculation 114 is performed in a Kalman filter, wherein a residual is formed between the measured pitch rate 110 and the estimated pitch rate. The estimated pitch dynamics 100a are corrected using the residual to obtain an updated pitch dynamics 100c.

[0037] In an exemplary embodiment, the uncertainty of the respective method steps is taken into account during the method flow. The uncertainty is represented by the respective covariance. The covariance is used to weight the individual values ​​when combining in the Kalman filter 112.

[0038] In an exemplary embodiment, a normal force 116 on an axle of the vehicle is derived from the updated pitch dynamics 100 c and used, for example, to adjust the braking force distribution.

[0039] Possible embodiments of the invention are subsequently summarized again or presented using slightly different wording.

[0040] An improved normal force estimation via an extended Kalman filter and a 6D sensor system is proposed.

[0041] Existing methods for estimating normal force are primarily based on the longitudinal acceleration detected for each vehicle. Estimation methods based solely on this acceleration typically use heuristics and characterizations of time characteristics. While these methods are tailored to the specific vehicle and simple to apply, they are imprecise. Advanced techniques use simple physical pitch models that simulate the wheel suspension and equate the pitch and yaw poles to the road surface. While these methods are based on physical principles and approximate realistic dynamics, their results are strongly influenced by model parameters and do not provide the desired accuracy when estimating normal force, particularly if they rely solely on acceleration-based input data.

[0042] The integration of 6D sensor technology now makes it possible to consider the pitch angular rate as a supplementary measured variable. This represents an additional important source of information for the pitch model. The implementation of an observer for data recording appears promising, as it allows the acquired data to be incorporated into the estimation. However, such an observer requires a very precise system model, preferably error-free signals, and a detailed configuration achieved by determining the static observer poles. In practice, these ideal conditions are often not met. Alternative methods for integrating the pitch angular rate are also difficult to implement, as sensor imperfections, measurement deviations, and signal drift can significantly affect the accuracy of the calculation.

[0043] The problem of inaccurate normal force estimation is effectively solved by the method proposed here. The traditional estimation method based mainly on longitudinal acceleration is significantly improved by integrating additional information from the pitch angular rate with the help of an extended 6D sensor system.

[0044] A key feature of the proposed method is its increased accuracy. The integration of the pitch angular rate as a complementary measured variable improves the estimation of the normal force. The combined system of the pitch model and the Kalman filter allows for the systematic consideration of measurement and model uncertainties. A further advantage is that the Kalman filter can compensate for potential inaccuracies in the model. This reduces the pressure to always maintain a perfectly parameterized model and increases the system's robustness against faults and deviations—a significant improvement compared to previous methods.

[0045] The Kalman filter is also characterized by its adaptability. It can dynamically adapt to changing conditions, which results in more precise estimates under changing environmental or driving conditions. Furthermore, its recursive structure enables real-time calculations, which is important in many applications.

[0046] In the method proposed here, an extended physical pitch model is combined with a partitioned Kalman filter that is supported by the integration of 6D sensor technology.

[0047] The expanded 6D sensor system is capable of detecting pitch angular rate. This additional information allows for more precise estimations based on the Kalman filter. The pitch model used takes into account longitudinal acceleration and, in addition, pitch dynamics. Previously, these pitch dynamics were only estimated, but can now be corrected using the additional sensor information. The Kalman filter serves as the optimal estimation algorithm for combining model predictions and measurement data. Due to its recursive structure, the Kalman filter is not only real-time-capable but also accounts for both measurement and model uncertainties. By partitioning the weighting matrix within the Kalman filter, the Kalman filter can be adapted to the system characteristics and optimally designed.

[0048] The estimation using the Kalman filter operates recursively and is divided into two main phases: a prediction phase and an update phase.

[0049] The prediction phase performs state prediction. Using a given pitch model, the state for the next time step is predicted. Furthermore, the prediction phase performs error covariance prediction. This phase calculates the uncertainty of the state prediction. This is reflected in the values ​​of the error covariance matrix. Larger values ​​indicate greater uncertainty in the prediction.

[0050] The residual calculation is performed in the update phase. The difference between the measured and predicted values ​​is calculated. This difference indicates how far the estimate deviates from the actual value and is used to correct the estimated value. In addition, the innovation covariance is determined in the update phase. The innovation covariance is a measure of the uncertainty about the residual. Here, the uncertainty of the prediction is combined with the uncertainty of the measurement. The Kalman gain is then determined in the update phase. This gain weights the measurement compared to the filter prediction. A high gain means a high trust in the measurement, a low value means a high trust in the prediction. In the state update in the update phase, the predicted state is updated using the Kalman gain, thereby obtaining an optimized state estimate. At the end of the update phase, an error covariance update is performed. The error covariance update takes into account the information from the current measurement and updates the uncertainty estimate of the system state.

[0051] This Kalman filter is characterized by a differentiated weighting operation using the uncertainty matrices of the measurement noise R and the process noise Q. Typically, specific values ​​are selected for these matrices, which are implemented in the form of diagonal matrices. In the method proposed here, however, the pitch angle rate provides the only state measurement considered in the Kalman filter. Furthermore, the model is not sufficiently precise in its estimation to be considered completely reliable. Therefore, specially adapted Q and R matrices are used. In this case, very high values ​​for the pitch angle and very low values ​​for the pitch angle rate are assumed in the R matrix. Conversely, high values ​​for the pitch angle rate and low values ​​for the pitch angle are used in the Q matrix. This differentiated weighting enables an optimized balance between process weighting and measurement weighting, resulting in an overall reliable estimate.

[0052] The normal force pre-control has a significant influence on the pitch angle rate. This can be detected in the measurement data, especially in the case of severe pitching in the vehicle. A significant change in the normal force pre-control occurs when the pitch angle rate measurement is omitted, i.e., when switching from a 6D to a 3D sensor system.

[0053] Finally, it is pointed out that terms such as "having", "comprising", etc. do not exclude other elements or steps, and terms such as "a" or "an" do not exclude a plurality. Reference signs in the claims are not to be considered as limiting.

Claims

1. A method for operating a vehicle, wherein: In a prediction phase (102), estimated pitch dynamics (100a) of the vehicle are predicted using the vehicle's current longitudinal acceleration (108) and a pitch model (106) of the vehicle, and in an update phase (104), updated pitch dynamics (100c) are determined using a partitioned Kalman filter (112), the estimated pitch dynamics (100a) of the vehicle, and a measured pitch angular rate (110).

2. The method according to claim 1, wherein The pitch angle rate (110) is measured using a 6D sensor system of the vehicle.

3. A method according to any one of the preceding claims, wherein In the updating phase (104), a residual is calculated using the estimated pitch dynamics (100a) and the measured pitch angular rate (110), wherein the estimated pitch dynamics (100a) is corrected using the residual in order to obtain the updated pitch dynamics (100c).

4. The method according to claim 3, wherein: An innovation covariance is calculated using the residual, wherein the updated pitch dynamics (100c) are determined using the innovation covariance.

5. A method according to any one of the preceding claims, wherein The covariance error of the estimated pitch dynamics (100a) is predicted using a previously determined covariance matrix, and an estimated covariance matrix is ​​estimated using the covariance error, wherein an updated covariance matrix is ​​determined using the estimated covariance matrix and the updated pitch dynamics (100c).

6. The method according to claim 5, wherein: Using the estimated covariance matrix, a Kalman gain for the estimated pitch dynamics (100a) and a Kalman gain for the measured pitch angle rate (110) are calculated, wherein the updated pitch dynamics (100c) are determined using the Kalman gains.

7. The method according to claim 6, wherein: The Kalman gain is calculated using an uncertainty matrix of the measurement noise of the measured pitch angular rate (110) and the process noise of the estimated pitch dynamics (100a).

8. A method according to any one of the preceding claims, wherein Using the updated pitching dynamics (100c), current normal forces (116) on the front and rear axles of the vehicle are determined.

9. A control unit, wherein: The control unit is designed to execute, implement and / or control the method according to one of the preceding claims in a corresponding device. 10 . A computer program product, which is designed to instruct a processor to execute, implement and / or control the method according to claim 1 when the computer program product is executed.

11. A machine-readable storage medium having stored thereon the computer program product according to claim 10.

Images