Vehicle inertia parameter estimation method based on intelligent tire and related device

Abstract

The present disclosure provides a kind of vehicle inertia parameter estimation method based on intelligent tire and related equipment, it is related to intelligent automobile state estimation and control technical field, especially in the field of vehicle inertia parameter estimation, the method comprises: through intelligent tire system direct output tire vertical force estimation value and tire lateral force estimation value, in the asynchronous information fusion of data information needing to use, after time domain alignment, again through the double architecture of the extended Kalman filter of load transfer model pre-construction and the Kalman filter of yaw dynamics model pre-construction realization joint estimation, obtain vehicle inertia parameter, without sufficient external excitation and additional model parameters, improve the online estimation precision of vehicle inertia parameter.

Description

Technical Field

[0001] This disclosure relates to the field of intelligent vehicle state estimation and control technology, particularly to the field of vehicle inertial parameter estimation, and especially to a method and related equipment for vehicle inertial parameter estimation based on intelligent tires. Background Technology

[0002] With the development of intelligent chassis technology, accurate perception and control of vehicle status has become an important foundation for achieving safety and intelligence. Vehicle inertial parameters (such as vehicle mass, center of gravity position, and yaw moment of inertia) have a decisive impact on vehicle dynamics performance, and their accuracy is directly related to the performance of vehicle state estimation and control systems.

[0003] However, in actual use, inertial parameters can change significantly due to factors such as loading conditions. Related technologies typically employ fixed values ​​or rely on onboard sensors and dynamic models for online estimation, but these methods often require sufficient external excitation to ensure convergence and are highly dependent on model accuracy.

[0004] Furthermore, due to the limited information available from vehicle-mounted sensors, the estimation of other inertial parameters, besides mass estimation, is constrained by the inability to directly obtain key observations such as vertical and lateral forces. This necessitates relying on additional model assumptions and introducing the state to be estimated, thereby introducing strong coupling between the state and the parameters, which limits the estimation accuracy and applicability.

[0005] It should be noted that the information disclosed in the background section above is only used to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] This disclosure provides a method and related equipment for estimating vehicle inertial parameters based on smart tires, which improves the accuracy and precision of vehicle inertial parameter estimation without requiring sufficient external excitation and additional model parameters.

[0007] Other features and advantages of this disclosure will become apparent from the following detailed description, or may be learned in part by practice of this disclosure.

[0008] In a first aspect, embodiments of this disclosure provide a method for estimating vehicle inertial parameters based on smart tires, characterized in that the method includes:

[0009] The intelligent tire system processes the acquired acceleration signals to obtain estimates of the tire vertical force and tire lateral force.

[0010] Asynchronous information fusion is performed on the longitudinal acceleration, lateral acceleration, and tire vertical force estimates in the chassis signal to obtain time-domain aligned longitudinal acceleration, lateral acceleration, and tire vertical force estimates.

[0011] The time-domain aligned longitudinal acceleration, lateral acceleration, and tire vertical force estimates are input into an extended Kalman filter pre-built using a load transfer model to obtain the first vehicle inertial parameters. The first vehicle inertial parameters include: vehicle mass estimate, centroid longitudinal position estimate, and centroid height estimate.

[0012] Asynchronous information fusion is performed on the yaw acceleration, tire lateral force estimate, and centroid longitudinal position estimate determined based on the yaw rate in the chassis signal to obtain time-domain aligned yaw acceleration, tire lateral force estimate, and centroid longitudinal position estimate.

[0013] The time-domain aligned yaw acceleration, tire lateral force estimates, and centroid longitudinal position estimates are input into a Kalman filter pre-built based on the yaw dynamics model to obtain the second vehicle inertial parameters; the second vehicle inertial parameters include: yaw moment of inertia estimates.

[0014] In one possible embodiment, the method further includes:

[0015] Taking one rotation of the smart tire as a cycle, sampling is performed at preset angles to obtain acceleration signals of the smart tire for multiple cycles; the acceleration signals include: longitudinal acceleration signal, lateral acceleration signal and vertical acceleration signal.

[0016] In one possible embodiment, the acquired acceleration signal is processed by an intelligent tire system to obtain estimates of the tire vertical force and the tire lateral force, including:

[0017] The first acceleration signal is obtained by taking the first acceleration signal of each tire as an independent sample from the acceleration signal.

[0018] The first acceleration signal is input into a low-pass filter to obtain the second acceleration signal;

[0019] Identify the ground imprint area within each tire rotation cycle and obtain the third acceleration signal from the second acceleration signal;

[0020] Based on the partial least squares method, acceleration feature points with a correlation greater than a threshold with tire force are extracted from the third acceleration signal;

[0021] The acceleration feature points are input into the trained Gaussian process regression model to obtain the tire vertical force estimate and the tire lateral force estimate.

[0022] In one possible embodiment, the longitudinal acceleration, lateral acceleration, and tire vertical force estimates in the chassis signal are asynchronously fused to obtain time-domain aligned estimates of longitudinal acceleration, lateral acceleration, and tire vertical force, including:

[0023] The first sliding buffer is constructed based on the timestamps of longitudinal and lateral acceleration;

[0024] A second sliding buffer is constructed based on the timestamp of the vertical force estimate for each tire in the tire vertical force estimate;

[0025] Obtain the estimated value of the first tire vertical force at the current recording time in the second sliding buffer; the estimated value of the first tire vertical force includes the estimated value of the vertical force of any tire or the estimated values ​​of the vertical force of two unrelated tires in the load transfer model;

[0026] The first interpolation time is determined based on the current recording time and the previous recording time corresponding to the vertical force estimate of each tire in the first tire vertical force estimate. The first interpolation time is the recording time with the smallest time difference between the current recording time and the previous recording time, where other tire vertical force estimates exist and are located between the current recording time and the previous recording time. The vertical force estimates of other tires are the vertical force estimates of tires that were not sampled at the current recording time among all tire vertical force estimates, and are the vertical force estimates of tires that form a measurement vector with the first tire vertical force estimate in the load transfer model. The previous recording time is the recording time of the first tire vertical force estimate before the current recording time in the second sliding buffer.

[0027] Determine the estimated value of the first tire vertical force at the first interpolation moment;

[0028] The estimated vertical force of the first tire, the estimated vertical force of other tires, and the longitudinal and lateral accelerations in the first sliding buffer at the first interpolation time are used as the estimated longitudinal acceleration, lateral acceleration, and tire vertical force for the current sub-time-domain alignment.

[0029] In one possible embodiment, the time-domain aligned estimates of longitudinal acceleration, lateral acceleration, and tire vertical force are input into an extended Kalman filter pre-built using a load transfer model to obtain the first vehicle inertial parameters, including:

[0030] Based on the load transfer model and the current tire vertical force estimate, the first measurement vector for the current iteration in the extended Kalman filter is constructed.

[0031] Obtain the predicted parameter values ​​and the first Kalman gain for the current update process in the extended Kalman filter;

[0032] Based on the parameter prediction values ​​of the current update process, the current longitudinal acceleration and lateral acceleration, determine the predicted observation values ​​for the current update.

[0033] Based on the parameter prediction values ​​of the current update process, the predicted observation values ​​of the current update, the first measurement vector of the current update, and the first Kalman gain of the current update, the parameter estimates of the current update are obtained; the parameter estimates are the first vehicle inertial parameters.

[0034] In one possible embodiment, the method further includes:

[0035] Determine the measurement Jacobian matrix and the first measurement noise covariance matrix in the extended Kalman filtering process; wherein, the first measurement noise covariance matrix is ​​set to a preset value or determined according to the confidence level corresponding to the tire vertical force estimate;

[0036] The observability assessment matrix is ​​determined based on the measurement Jacobian matrix and the first measurement noise covariance matrix.

[0037] Based on the main diagonal elements of the observability evaluation matrix, evaluate the first vehicle inertial parameters for the current iteration.

[0038] If any of the main diagonal elements is less than the first preset threshold, the update process of the first vehicle inertia parameter corresponding to any of the main diagonal elements is paused, and the first vehicle inertia parameter corresponding to any of the main diagonal elements in the previous iteration is used as the first vehicle inertia parameter corresponding to any of the main diagonal elements in the current iteration.

[0039] In one possible embodiment, the yaw acceleration, tire lateral force estimate, and center of gravity longitudinal position estimate determined based on the yaw rate in the chassis signal are asynchronously fused to obtain time-domain aligned yaw acceleration, tire lateral force estimate, and center of gravity longitudinal position estimate, including:

[0040] Determine the yaw rate acceleration based on the yaw rate in the chassis signal;

[0041] A third sliding buffer is constructed based on the timestamp of the yaw angle acceleration;

[0042] A fourth sliding buffer is constructed based on the timestamp of the lateral force estimate for each tire in the tire lateral force estimate;

[0043] A fifth sliding buffer is constructed based on the timestamps of the centroid longitudinal position estimates; wherein, the sampling frequency of the fifth sliding buffer is the same as that of the third sliding buffer;

[0044] Obtain the first tire lateral force estimate at the current recording time in the fourth sliding buffer; the first tire lateral force estimate includes the lateral force estimate of at least one tire in the yaw dynamics model; the number of at least one tire is less than 4;

[0045] Get any tire lateral force estimate that has not been updated at the current recording time;

[0046] The second interpolation time is determined based on the current recording time and the previous recording time corresponding to the lateral force estimate of each tire in the first tire lateral force estimate, as well as any tire lateral force estimate that has not been updated. The second interpolation time is the recording time with the smallest time difference between the current recording time and the previous recording time, and is located between two recording times of other tire lateral force estimates that have not been updated in the yaw dynamics model. The previous recording time is the recording time of the first tire lateral force estimate that was recorded before the current recording time in the fourth sliding buffer. Other tire lateral force estimates are tire lateral force estimates that have not been updated except for any tire lateral force estimate that has not been updated.

[0047] At the second interpolation time, the estimated lateral force for each tire is determined;

[0048] The yaw acceleration at the second interpolation time in the third sliding buffer, the estimated lateral force of each tire at the second interpolation time in the fourth sliding buffer, and the estimated longitudinal position of the centroid at the second interpolation time in the fifth sliding buffer are used as the yaw acceleration, tire lateral force, and centroid longitudinal position estimates for the current time-domain alignment.

[0049] In one possible embodiment, the time-domain aligned yaw rate acceleration, tire lateral force estimates, and centroid longitudinal position estimates are input into a Kalman filter pre-built using the yaw dynamics model to obtain second vehicle inertial parameters, including:

[0050] Based on the yaw dynamics model and the estimated tire lateral force, a second measurement vector using Kalman filtering is constructed.

[0051] Obtain the predicted parameter values ​​and the second Kalman gain for the current update process in the Kalman filter;

[0052] Construct a measurement matrix based on the current yaw acceleration;

[0053] Based on the measurement matrix, the second measurement vector, the parameter prediction value of the current update process, and the second Kalman gain of the current update, the parameter estimate value of the current update is determined. The parameter estimate value is the second vehicle inertial parameter.

[0054] In one possible embodiment, the method further includes:

[0055] Determine the current measurement matrix and the second measurement noise covariance matrix in the Kalman filtering process; wherein the second measurement noise covariance matrix is ​​set to a preset value or determined according to the confidence level corresponding to the tire lateral force estimate and the centroid longitudinal position estimate;

[0056] Based on the current measurement matrix and the second measurement noise covariance matrix, determine the observability assessment value;

[0057] Based on the observability assessment value, evaluate the second vehicle inertial parameters for the current test.

[0058] If the observability assessment value is less than the second preset threshold, the update process of the second vehicle inertia parameter is paused, and the previous second vehicle inertia parameter is used as the current second vehicle inertia parameter.

[0059] Secondly, embodiments of this disclosure provide a vehicle inertial parameter estimation device based on smart tires, comprising:

[0060] The first estimation unit is used to process the acquired acceleration signal through the intelligent tire system to obtain the estimated values ​​of the tire vertical force and the tire lateral force.

[0061] The first alignment unit asynchronously fuses the longitudinal acceleration, lateral acceleration, and tire vertical force estimates from the chassis signal to obtain time-domain aligned longitudinal acceleration, lateral acceleration, and tire vertical force estimates.

[0062] The second estimation unit is used to input the time-domain aligned longitudinal acceleration, lateral acceleration, and tire vertical force estimates into an extended Kalman filter pre-built based on a load transfer model to obtain the first vehicle inertial parameters. The first vehicle inertial parameters include: vehicle mass estimate, centroid longitudinal position estimate, and centroid height estimate.

[0063] The second alignment unit is used to asynchronously fuse the yaw acceleration, tire lateral force estimate, and centroid longitudinal position estimate determined based on the yaw rate in the chassis signal to obtain time-domain aligned yaw acceleration, tire lateral force estimate, and centroid longitudinal position estimate.

[0064] The third estimation unit is used to input the time-domain aligned yaw acceleration, tire lateral force estimate, and centroid longitudinal position estimate into a Kalman filter pre-built based on the yaw dynamics model to obtain the second vehicle inertial parameters; the second vehicle inertial parameters include: yaw moment of inertia estimate.

[0065] In one possible embodiment, it further includes:

[0066] The first observability evaluation unit is used to determine the measurement Jacobian matrix and the first measurement noise covariance matrix in the extended Kalman filtering process; wherein, the first measurement noise covariance matrix is ​​set to a preset value or determined according to the confidence level corresponding to the tire vertical force estimate;

[0067] The observability assessment matrix is ​​determined based on the measurement Jacobian matrix and the first measurement noise covariance matrix.

[0068] Based on the main diagonal elements of the observability evaluation matrix, evaluate the first vehicle inertial parameters for the current iteration.

[0069] If any of the main diagonal elements is less than the first preset threshold, the update process of the first vehicle inertia parameter corresponding to any of the main diagonal elements is paused, and the first vehicle inertia parameter corresponding to any of the main diagonal elements in the previous iteration is used as the first vehicle inertia parameter corresponding to any of the main diagonal elements in the current iteration.

[0070] In one possible embodiment, it further includes:

[0071] The second observability evaluation unit is used to determine the current measurement matrix and the second measurement noise covariance matrix in the Kalman filtering process; wherein, the second measurement noise covariance matrix is ​​set to a preset value or determined according to the confidence level corresponding to the tire lateral force estimate and the centroid longitudinal position estimate.

[0072] Based on the current measurement matrix and the second measurement noise covariance matrix, determine the observability assessment value;

[0073] Based on the observability assessment value, evaluate the second vehicle inertial parameters for the current test.

[0074] If the observability assessment value is less than the second preset threshold, the update process of the second vehicle inertia parameter is paused, and the previous second vehicle inertia parameter is used as the current second vehicle inertia parameter.

[0075] Thirdly, embodiments of this disclosure provide an electronic device, including: a processor; and a memory for storing executable instructions of the processor; wherein the processor is configured to perform the method of the first aspect described above by executing the executable instructions.

[0076] Fourthly, embodiments of this disclosure provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect above.

[0077] Fifthly, according to another aspect of this disclosure, a computer program product or computer program is also provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and the processor executes the computer instructions, causing the computer device to perform any of the methods described above.

[0078] This disclosure provides a method and related equipment for estimating vehicle inertial parameters based on intelligent tires, relating to the field of intelligent vehicle state estimation and control technology, and particularly to the field of vehicle inertial parameter estimation. The method includes: directly outputting tire vertical force estimates and tire lateral force estimates through an intelligent tire system; asynchronously fusing longitudinal acceleration, lateral acceleration, and tire vertical force estimates from chassis signals; inputting the time-domain aligned data into an extended Kalman filter pre-built using a load transfer model to obtain first vehicle inertial parameters; and asynchronously fusing yaw acceleration, tire lateral force estimates, and centroid longitudinal position estimates determined based on yaw rate from chassis signals; inputting the time-domain aligned data into a Kalman filter pre-built using a yaw dynamics model to obtain second vehicle inertial parameters. This method can improve the accuracy and precision of online estimation of vehicle inertial parameters without requiring sufficient external excitation and additional model parameters.

[0079] It should be understood that the above general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Attached Figure Description

[0080] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure. It is obvious that the drawings described below are merely some embodiments of this disclosure, and those skilled in the art can obtain other drawings based on these drawings without any inventive effort.

[0081] Figure 1 A flowchart of a vehicle inertial parameter estimation method based on smart tires is shown in an embodiment of this disclosure;

[0082] Figure 2 This diagram illustrates a flowchart of an embodiment of the present disclosure for obtaining a tire force estimate.

[0083] Figure 3 This diagram illustrates the lateral force under a serpentine working condition according to an embodiment of the present disclosure.

[0084] Figure 4 This diagram illustrates the vertical force under a serpentine working condition in an embodiment of the present disclosure.

[0085] Figure 5 A schematic diagram of a lateral load transfer model according to an embodiment of this disclosure is shown;

[0086] Figure 6 A schematic diagram of a longitudinal load transfer model according to an embodiment of this disclosure is shown;

[0087] Figure 7 A schematic diagram of a vehicle yaw dynamics model according to an embodiment of this disclosure is shown;

[0088] Figure 8 This diagram illustrates a flowchart of an asynchronous information fusion method according to an embodiment of the present disclosure;

[0089] Figure 9 A schematic diagram illustrating an asynchronous information fusion embodiment of this disclosure is shown;

[0090] Figure 10 A flowchart illustrating the observability evaluation of an extended Kalman filter according to an embodiment of this disclosure is shown;

[0091] Figure 11 This diagram illustrates a flowchart of an extended Kalman filter process according to an embodiment of the present disclosure;

[0092] Figure 12 This diagram illustrates a flowchart of another asynchronous information fusion method according to an embodiment of the present disclosure;

[0093] Figure 13 This diagram illustrates a Kalman filtering process according to an embodiment of the present disclosure.

[0094] Figure 14 A flowchart illustrating the observability evaluation of a Kalman filter according to an embodiment of this disclosure is shown;

[0095] Figure 15 A schematic diagram showing the effect of yaw acceleration estimation in an embodiment of this disclosure is provided.

[0096] Figure 16 This diagram illustrates the estimation results of vehicle inertia parameters under a simulated dual lane change condition according to an embodiment of the present disclosure.

[0097] Figure 17 A schematic diagram showing the estimated value of tire vertical force under a double lane change condition in an embodiment of this disclosure is provided.

[0098] Figure 18 This diagram illustrates the tire lateral force estimation under a double lane change condition in an embodiment of the present disclosure.

[0099] Figure 19 A schematic diagram showing the inertial parameter identification results of a real vehicle under double lane change conditions in an embodiment of this disclosure;

[0100] Figure 20 This diagram illustrates the structure of a vehicle inertial parameter estimation device based on smart tires, according to an embodiment of this disclosure.

[0101] Figure 21 A schematic diagram of the structure of an electronic device according to an embodiment of the present disclosure is shown. Detailed Implementation

[0102] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that this disclosure will be more comprehensive and complete, and will fully convey the concept of the exemplary embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0103] Furthermore, the accompanying drawings are merely illustrative of this disclosure and are not necessarily drawn to scale. The same reference numerals in the drawings denote the same or similar parts, and therefore repeated descriptions of them will be omitted. Some block diagrams shown in the drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities may be implemented in software, in one or more hardware modules or integrated circuits, or in different network and / or processor devices and / or microcontroller devices.

[0104] The following detailed description of this exemplary implementation method is provided in conjunction with the accompanying drawings and embodiments.

[0105] First, this disclosure provides a method for estimating vehicle inertial parameters based on smart tires. This method can be executed by any electronic device with computing capabilities, such as a terminal device or a server. In the following process, an electronic device is used as an example of a terminal device.

[0106] Figure 1 This diagram illustrates a method for estimating vehicle inertial parameters based on smart tires, as described in an embodiment of this disclosure. Figure 1 As shown in the embodiments of this disclosure, the method for estimating vehicle inertial parameters based on smart tires includes the following steps:

[0107] S102: The acquired acceleration signal is processed by the intelligent tire system to obtain the estimated values ​​of the tire vertical force and the tire lateral force.

[0108] In one possible embodiment, related technologies estimate tire forces based on known vehicle inertial parameters. However, in this application, an intelligent tire system can directly estimate the vertical and lateral tire forces without requiring additional longitudinal or lateral excitation conditions.

[0109] In one possible embodiment, the acceleration signal in S102 can be obtained by taking one rotation of the smart tire as a cycle and sampling at a preset angle as the sampling interval to obtain the acceleration signal of the smart tire for multiple cycles; the acceleration signal includes: longitudinal acceleration signal, lateral acceleration signal and vertical acceleration signal.

[0110] For example, a triaxial (longitudinal, lateral and vertical) acceleration sensor is installed inside the smart tire. The acceleration sensor collects acceleration signals according to the rotation cycle of each tire. One rotation cycle is divided into 360 degrees. The preset angle can be set to 1 degree or 5 degrees. In this disclosure, 5 degrees is used as an example.

[0111] Figure 2 A flowchart illustrating an embodiment of this disclosure for obtaining a tire force estimate is shown, as follows: Figure 2 As shown, it includes the following steps:

[0112] S202: The acceleration signal is obtained by taking the first acceleration signal of each tire as an independent sample.

[0113] S204: Input the first acceleration signal into the low-pass filter to obtain the second acceleration signal.

[0114] S206: Identify the ground imprint area during each tire rotation cycle and obtain the third acceleration signal from the second acceleration signal.

[0115] S208: Extract acceleration feature points in the third acceleration signal that are correlated with tire force greater than a threshold based on partial least squares method.

[0116] S210: Input the acceleration feature points into the trained Gaussian process regression model to obtain the tire vertical force estimate and tire lateral force estimate.

[0117] In one possible embodiment, the acceleration signal of each tire's single rotation cycle is divided into independent samples and input into a low-pass filter to filter out high-frequency noise. By identifying the ground imprint region, the third acceleration signal corresponding to the ground imprint region is extracted from the second acceleration signal. The third acceleration signal is then processed, and acceleration feature points with a correlation greater than a threshold with the tire force are extracted from the third acceleration signal based on the partial least squares method as features used to determine the tire vertical force estimate and the tire lateral force estimate.

[0118] The acceleration feature points are input into a trained Gaussian Process Regression (GPR) model to complete the estimation process and obtain the estimated values ​​of the tire vertical force and the tire lateral force.

[0119] The above process mainly uses acceleration signals. By extracting data with high correlation to tire force as feature points to be processed, tire force can be estimated. Not only can the estimated values ​​of tire vertical force and tire lateral force be obtained without sufficient external excitation, but the accuracy of the obtained estimated values ​​of tire vertical force and tire lateral force can also be improved based on the highly correlated feature points.

[0120] In one possible embodiment, real-vehicle data was collected at a test track to train the Gaussian process regression model. The collected data included acceleration signals and tire forces from all four tires, as well as chassis signals from the vehicle, including signals from the inertial measurement unit (IMU) and steering wheel angle.

[0121] The data collection conditions included constant speed straight driving, straight acceleration and deceleration, serpentine maneuvers, and double lane changeovers, as detailed in Table 1. The constant speed straight driving condition involved driving in a straight line at a stable speed of 30, 40, 60, 70, and 90 km / h; the straight acceleration and deceleration condition involved the vehicle accelerating from a standstill to 100 km / h and then braking to a stop; the serpentine and double lane changeover conditions were tested at 30, 45, and 60 km / h. Table 1 shows the actual vehicle test conditions.

[0122] Table 1

[0123]

[0124] During model training, the training set, validation set, and test set can be divided in various ways, and this disclosure does not limit the methods used.

[0125] For example, the 60 km / h constant speed straight driving and double lane change scenarios can be used as unfamiliar data for algorithm validation, while the remaining scenario data can be used for training. The ratio of the training set, validation set, and test set is 0.6, 0.2, and 0.2, respectively.

[0126] The error statistics of the four tires under different working conditions are shown in Tables 2 and 3. Table 2 is the statistical table of lateral force estimation error, and Table 3 is the statistical table of vertical force estimation error. Tables 2 and 3 respectively list the mean error (ME), mean absolute error (MAE), and normalized root mean square error (NRMSE).

[0127] Table 2

[0128]

[0129] Table 3

[0130]

[0131] The results in the table above show that the NRMSE for lateral force estimation is less than 5%, and the NRMSE for vertical force estimation is less than 2%, indicating high overall accuracy. The average error for each tire is close to zero, suggesting that there is no significant systematic bias in the prediction results.

[0132] Furthermore, Figure 3 A schematic diagram of lateral forces under a serpentine working condition in an embodiment of this disclosure is shown, as follows: Figure 3 As shown, the vertical axis represents tire lateral force, and the horizontal axis represents the sample number, including: the estimated value, the measured value, and the 95% confidence interval of the tire lateral force. Figure 3 The 95% confidence interval is represented by a grayscale range. Solid lines represent measured values, and dashed lines represent estimated values. For clarity, the measured and estimated values ​​are shown in two coordinate systems.

[0133] Figure 3 In the coordinate system, the vertical axis includes the coordinates -4, -2, 0, and 2, and the horizontal axis includes the coordinates 0, 10, 20, 30, 40, and 50.

[0134] Figure 4 A schematic diagram of vertical force under a serpentine working condition in an embodiment of this disclosure is shown, as follows: Figure 4 As shown, the vertical axis represents the tire vertical force, and the horizontal axis represents the sample number, including: the estimated value, the measured value, and the 95% confidence interval of the tire vertical force. Figure 3 The 95% confidence interval is represented by grayscale ranges, with solid lines representing measured values ​​and dashed lines representing estimated values. For clarity, the measured and estimated values ​​are shown in two coordinate systems.

[0135] Figure 4 In the coordinate system, the vertical axis includes the coordinates -4, -2, 2, and 10, and the horizontal axis includes the coordinates 0, 10, 20, 30, 40, and 50.

[0136] pass Figure 3 and Figure 4 It can be seen that the measured and estimated values ​​of tire lateral force, as well as the estimated value of tire vertical force, are within the 95% confidence interval, indicating that the accuracy and precision of the estimated values ​​are relatively high and basically match the measured values. The estimated values ​​of tire force can be used in turn to estimate the inertial parameters of the whole vehicle.

[0137] S104: Asynchronously fuse the longitudinal acceleration, lateral acceleration, and tire vertical force estimates in the chassis signal to obtain time-domain aligned longitudinal acceleration, lateral acceleration, and tire vertical force estimates.

[0138] In one possible embodiment, the time-domain aligned longitudinal acceleration, lateral acceleration, and tire vertical force estimates represent alignment at the time level, wherein at least in the extended Kalman filtering process of the load transfer model, the vertical force estimates of the two related tires constituting a measurement vector are aligned in the time domain for each tire.

[0139] S106: Input the time-domain aligned longitudinal acceleration, lateral acceleration, and tire vertical force estimates into the extended Kalman filter pre-built using the load transfer model to obtain the first vehicle inertial parameters. The first vehicle inertial parameters include: vehicle mass estimate, centroid longitudinal position estimate, and centroid height estimate.

[0140] In one possible embodiment, the load transfer model may include a lateral load transfer model and a longitudinal load transfer model.

[0141] Figure 5 A schematic diagram of a lateral load transfer according to an embodiment of this disclosure is shown, such as... Figure 5 As shown, this includes: CG (Center of Gravity) represents the vehicle's center of gravity. The lateral acceleration is represented by t, the wheelbase is represented by h, and the center of gravity height is represented by h. This represents the vertical force on the right front tire. This represents the vertical force on the left rear tire. This represents the vertical force on the left front tire. This represents the vertical force on the right rear tire.

[0142] Figure 6 A schematic diagram of a longitudinal load transfer according to an embodiment of this disclosure is shown, such as... Figure 6 As shown, CG represents the vehicle's center of gravity. This represents longitudinal acceleration, and L represents wheelbase. This represents the vertical force on the right front tire. This represents the vertical force on the left rear tire. This represents the vertical force on the left front tire. This represents the vertical force on the right rear tire, where 'a' is the distance from the center of mass to the front axle, i.e., the longitudinal position of the center of mass.

[0143] pass Figure 5 and Figure 6 This describes the relationship between the vertical force of the tires and the longitudinal and lateral accelerations at the center of gravity, given the vehicle's geometry and overall inertial parameters. The two models mentioned above assume the vehicle is a rigid body, neglecting the pitch, roll, and vertical motion of the sprung mass caused by suspension compression, as well as the lateral shift of the vehicle's center of gravity.

[0144] After considering the static mass distribution and the load transfer caused by inertial forces, the vertical force of the tire can be expressed in the following forms, including formulas (1) to (4):

[0145] (1)

[0146] (2)

[0147] (3)

[0148] (4)

[0149] in, Indicates the overall vehicle weight. It represents the acceleration due to gravity.

[0150] The above Figure 5 and Figure 6 The parameters involved in the corresponding formulas (1)-(4) represent actual or measured values.

[0151] In one possible embodiment, the extended Kalman filter can be constructed using a load transfer model and a random walk model, and the overall processing procedure of the extended Kalman filter can be defined. The inputs to the extended Kalman filter are the estimated vertical force, lateral acceleration, and longitudinal acceleration of each tire. The dynamic process of the parameters is modeled using a random walk model, and the first state equations corresponding to the extended Kalman filter system are shown in Equations (5) and (6) below.

[0152] (5)

[0153] (6)

[0154] in, This represents the parameter estimate for the k-th iteration of the extended Kalman filter. This represents the parameter estimates for the (k-1)th iteration of the extended Kalman filter. This represents the process noise of the (k-1)th iteration of the extended Kalman filter, where it can be expressed as... Let N represent a Gaussian distribution. This represents the first-process covariance matrix of the extended Kalman filter. The measurement noise of the k-th iteration of the extended Kalman filter can be represented by... Let R be the first measurement noise covariance matrix of the extended Kalman filter. Let represent the measurement equation for the k-th iteration of the extended Kalman filter, where the calculated value of the measurement equation is the predicted observation when the input is explicitly stated in the measurement equation. For example, for the above formula (6), when there is an explicit input, At that time, The corresponding calculated value is the predicted observation value for the kth time. Let A represent the first measurement vector of the k-th iteration of the extended Kalman filter, and let A denote the identity matrix.

[0155] An extended Kalman filter is constructed using a load transfer model, where the parameter estimates and the first measurement vector for each output can be constructed as shown in Equations (7) and (8) below.

[0156] (7)

[0157] (8)

[0158] in, This represents the estimated vehicle weight. This represents the estimated longitudinal position of the centroid. This represents the estimated height of the centroid. This represents the estimated vertical force on the right front tire. This represents the estimated vertical force on the left rear tire. This represents the estimated vertical force on the left front tire. This represents the estimated vertical force on the right rear tire.

[0159] x represents the parameter estimates used to construct the extended Kalman filter definition, and z represents the first measurement vector used to construct the extended Kalman filter definition.

[0160] Formulas (7) and (8) above do not represent specific parameters of a certain processing procedure, but rather parameters defined during the construction of the extended Kalman filter.

[0161] By combining the above formulas (7) and (8) with the load transfer model, the parameter matrix defined by the measurement equation for each step in the extended Kalman filter process can be constructed as shown in the following formula (9).

[0162] (9)

[0163] in, The measurement equation is represented by formula (9), and the parameters involved in formula (9) are described in the previous formulas, so they will not be repeated here. L and t correspond to the actual values. and This represents the longitudinal and lateral acceleration measured from chassis signals. The problem to be solved is... This represents the estimated vehicle weight. This represents the estimated longitudinal position of the centroid. This represents the estimated height of the centroid.

[0164] The first vehicle inertial parameters are obtained by estimating the matrix constructed above using the extended Kalman filter process.

[0165] As can be seen from the above formula, obtaining the estimated vertical force of a tire requires more than just one estimate; it necessitates inputting the vertical force estimates of two related tires that form a measurement vector within the load transfer model as a single data set. However, due to differences in the installation position and wheel speed of the intelligent tire system, the update time of each tire's vertical force estimate will vary. Furthermore, the intelligent tire force estimation results may exhibit inconsistent output frequency and asynchronous updates across the four tires. Therefore, the vertical force of different tires needs to be analyzed in real-time. Domain alignment is crucial because the update frequency of the intelligent tire system varies with wheel speed and is typically lower than the chassis signal frequency. If the hold-and-add method is directly used as in related technologies, significant hysteresis errors can easily be introduced. Therefore, for S104, it is necessary to align the vertical force estimates of the two related tires that constitute a measurement vector in the time domain, and also to align the tire vertical force estimates with the chassis signal in the time domain before performing subsequent calculations. This improves the consistency of the data input to the extended Kalman filter process and achieves effective fusion of asynchronous measurements through asynchronous information fusion, thereby improving the accuracy of estimating the vehicle's inertial parameters.

[0166] S108: Asynchronously fuse the yaw acceleration, tire lateral force estimate, and centroid longitudinal position estimate determined based on the yaw rate in the chassis signal to obtain time-domain aligned yaw acceleration, tire lateral force estimate, and centroid longitudinal position estimate.

[0167] In one possible implementation, the time-domain aligned yaw acceleration, tire lateral force estimates, and centroid longitudinal position estimates represent alignment at the time level. The Kalman filter process for constructing the yaw dynamics model requires time-domain alignment of all tire lateral force estimates.

[0168] S110: Input the time-domain aligned yaw acceleration, tire lateral force estimate, and centroid longitudinal position estimate into a Kalman filter pre-built based on the yaw dynamics model to obtain the second vehicle inertial parameters, which include the yaw moment of inertia estimate.

[0169] In one possible embodiment, Figure 7 A schematic diagram of a vehicle yaw dynamics model according to an embodiment of this disclosure is shown, such as... Figure 7 As shown, it includes: This indicates the lateral force on the right front tire. This indicates the lateral force on the left rear tire. This indicates the lateral force on the left front tire. The value represents the lateral force of the right rear tire, CG represents the vehicle's center of gravity, L represents the wheelbase, r represents the yaw rate, t represents the track width, and a is the distance from the center of gravity to the front axle, i.e., the longitudinal position of the center of gravity. This refers to the steering angle of the front wheels.

[0170] Since the estimation of yaw moment of inertia can be completed under normal turning conditions, which typically involve relatively small longitudinal forces, the longitudinal forces of the tires can be ignored during the modeling process in this embodiment of the disclosure.

[0171] The process of solving the yaw moment of inertia is expressed by the following formula (10). As shown in formula (10), the front wheel angle is ignored in the following formula because the front wheel angle is generally small.

[0172] (10)

[0173] in, This means that the derivative of the yaw rate is the yaw acceleration. This represents the moment of inertia during yaw. (The above...) Figure 7 The parameters involved in the corresponding formula (10) represent actual or measured values.

[0174] In this embodiment, the inertial parameters of the second vehicle are estimated using Kalman filtering, and the data in the Kalman filtering process is constructed from the yaw dynamics model.

[0175] The second state equation corresponding to the Kalman filtering process is as follows, as shown in equations (11) and (12).

[0176] (11)

[0177] (12)

[0178] in, This represents the parameter estimate for the k-th iteration of the Kalman filter. This represents the parameter estimate for the (k-1)th iteration of the Kalman filter. The process noise of the (k-1)th Kalman filter can be represented by... Let N represent a Gaussian distribution. This represents the covariance matrix of the second process in Kalman filtering. The measurement noise of the k-th Kalman filter can be represented by... express, This is the second measurement noise covariance matrix of the Kalman filter. This represents the measurement matrix for the k-th Kalman filter. Let A represent the second measurement vector of the k-th Kalman filter, and let A denote the identity matrix.

[0179] The parameter estimates, measurement matrix, and second measurement vector for each output can be constructed in the following ways, as shown in formulas (13)-(15).

[0180] (13)

[0181] in, This represents the parameter estimates used to construct the Kalman filter definition. This represents the estimated value of the yaw moment of inertia.

[0182] (14)

[0183] in, This indicates the construction of the measurement matrix defined by the Kalman filter.

[0184] (15)

[0185] in, This represents the second measurement vector defined by the Kalman filter. This represents the estimated lateral force on the left front tire. This represents the estimated lateral force on the right front tire. This represents the centroid longitudinal position estimate from the extended Kalman filter output. This represents the estimated lateral force on the left rear tire. This represents the estimated lateral force of the right rear tire.

[0186] The second vehicle inertial parameter is obtained by estimating the matrix constructed above using a Kalman filter process.

[0187] The method described in this embodiment provides highly accurate estimates of the vertical and lateral forces of the tires obtained directly from the intelligent tire system. Due to the difference between the sampling frequency and the output frequency of the intelligent tire system, these estimates are used for the estimation of the vehicle's inertial parameters after time-domain alignment. The estimation process utilizes a dual architecture of extended Kalman filtering (built by the load transfer model) and Kalman filtering (built by the yaw dynamics model) to obtain the vehicle's inertial parameters. Further time-domain alignment is required between the extended Kalman filtering and the Kalman filtering. The data used in the overall process is aligned in the time domain, and the tire force data used in the overall estimation process is directly estimated from the acceleration signal. This avoids the cumulative calculation errors caused by redundant calculations in related technologies for estimating vehicle inertial parameters.

[0188] Regarding S104, Figure 8 A flowchart of an asynchronous information fusion embodiment of this disclosure is shown, such as... Figure 8 As shown, it includes the following steps:

[0189] S802: Construct the first sliding buffer based on the timestamps of longitudinal and lateral acceleration.

[0190] S804: Construct a second sliding buffer based on the timestamp of the vertical force estimate for each tire in the tire vertical force estimate.

[0191] S806: Obtain the estimated value of the first tire vertical force at the current recording time in the second sliding buffer. The estimated value of the first tire vertical force includes the estimated value of the vertical force of any one tire or the estimated values ​​of the vertical force of two unrelated tires in the load transfer model.

[0192] S808: Determine the first interpolation time based on the current recording time corresponding to the vertical force estimate of each tire in the first tire vertical force estimate and the previous recording time of the current recording time.

[0193] The first interpolation time is the estimated vertical force of other tires, located between the current recording time and the previous recording time, and the recording time with the smallest time difference from the current recording time.

[0194] Among them, the vertical force estimates of other tires are the vertical force estimates of tires that were not sampled at the current recording time among all the vertical force estimates of tires, and are the vertical force estimates of tires that form a measurement vector with the vertical force estimate of the first tire in the load transfer model.

[0195] The previous recording time is the time when the estimated value of the first tire vertical force was recorded before the current recording time in the second sliding buffer.

[0196] In one possible embodiment, Figure 9A schematic diagram of asynchronous information fusion in an embodiment of this disclosure is shown, such as... Figure 9 As shown, a sliding buffer is included. The data sources include FL, FR, RL, and RR, representing the left front wheel, right front wheel, left rear wheel, and right rear wheel of the vehicle, respectively. The corresponding sliding buffer is the second sliding buffer. The sliding buffer corresponding to the chassis signal is the first sliding buffer. The data stream is recorded in the form of timestamps, where 1 indicates that data has been updated in the sliding buffer, and 0 indicates that data has not been updated. The current time and the first interpolation time are... Figure 9 As shown in the figure, the last time point is used as the current time point, and the first interpolation time point is selected. Figure 9 The time shown in the image Figure 9 The previous recording time is shown in the image, such as... Figure 9 As shown. In order to reduce the use of computing power and storage capacity in the asynchronous information fusion process, the data of the two most recent moments can be selected to be retained in the second sliding buffer. Since there are 4 tires, the data of 8 moments can be selected to be retained in the first sliding buffer. This maximizes the ability to align in the time domain and saves computing power and storage capacity.

[0197] Taking the current moment as an example with the vertical force estimate of one tire, and specifically the vertical force estimate of the left front tire, when selecting the first interpolation moment, the data obtained during interpolation between the previous recording moment and the current moment is more accurate and stable. The vertical force estimate that can form a measurement vector (i.e., the first measurement vector) with the left front tire in the load transfer model is selected. This means choosing the vertical force estimate corresponding to FR and the vertical force estimate corresponding to RL. There are two options. The time difference between the recording moment and the current moment for each of these two vertical force estimates is calculated. The recording moment with the smallest time difference from the current moment is then selected. Figure 9 The first interpolation time shown is chosen so that the estimated value of the first tire vertical force obtained at the first interpolation time is closest to the estimated value of the first tire vertical force at the current time, and is more accurate.

[0198] In one possible embodiment, if the second sliding buffer retains P time points, where P is a positive integer greater than 2, then the first interpolation time point is at least between the current time point and the earliest time point where updated data exists.

[0199] In one possible embodiment, if at least two tire vertical force estimates that can constitute a first measurement vector are obtained at the current recording time, the data at the current time can be directly obtained as time-domain aligned longitudinal acceleration, lateral acceleration, and tire vertical force estimates.

[0200] For example, if the vertical force estimates of the three tires are obtained at the current moment, the subsequent extended Kalman filtering process can be executed directly.

[0201] For example, if the vertical force estimates corresponding to FL and RL are obtained at the current moment, the subsequent extended Kalman filtering process can be executed directly.

[0202] For example, if the vertical force estimates for the four tires are obtained at the current moment, the subsequent extended Kalman filtering process can be executed directly.

[0203] S810: Determine the estimated value of the first tire vertical force at the first interpolation moment.

[0204] In one possible embodiment, the vertical force estimate of FL at the first interpolation time is estimated as the first tire vertical force estimate based on the first tire vertical force estimate recorded at the previous recording time and the current time, i.e., the vertical force estimate of FL at these two times.

[0205] This embodiment uses interpolation to determine the interpolation time and corresponding data. The estimation in this embodiment is online; the data is generated and sampled in real time, and the estimated tire force is also processed and output in real time. Using extrapolation results in lower data accuracy compared to the interpolation method used in this embodiment.

[0206] S812: The estimated vertical force of the first tire, the estimated vertical force of other tires, the longitudinal acceleration and lateral acceleration in the first sliding buffer at the first interpolation time are used as the estimated longitudinal acceleration, lateral acceleration and tire vertical force of the current sub-time domain alignment.

[0207] In one possible embodiment, the time-domain aligned longitudinal acceleration, lateral acceleration, and tire vertical force estimates obtained each time can be input into a subsequent extended Kalman filter process to obtain estimated parameters.

[0208] It should be noted that the tire vertical force estimate obtained from each time-domain alignment does not necessarily include the tire vertical force estimates of all four tires. Only a set of first measurement vectors is needed for one estimation. During the estimation process, the first update flag of the measurement equation is defined. This indicates whether the estimated tire vertical force has been updated. The first update flag includes four data points, and the specific values ​​of the elements in the first update flag are related to... The corresponding measurement vectors are defined as follows: if the i-th measurement vector has not been updated, the i-th element in the first update flag is 0; if the i-th measurement vector has been updated, the i-th element in the first update flag is 1. This represents a 1x4 matrix. During the process of updating the first Kalman gain in the extended Kalman filter, the first update flag indicates whether the first measurement vector has been updated, thus completing the estimation process of the extended Kalman filter.

[0209] In one possible embodiment, after asynchronous information fusion is completed based on the above process, an extended Kalman filter process is then performed to perform observability evaluation.

[0210] Figure 10 A flowchart illustrating the observability evaluation of an extended Kalman filter according to an embodiment of this disclosure is shown, such as... Figure 10 As shown, it includes the following steps:

[0211] S1002: Determine the measurement Jacobian matrix and the first measurement noise covariance matrix in the extended Kalman filtering process, wherein the first measurement noise covariance matrix is ​​set to a preset value or determined according to the confidence level corresponding to the tire vertical force estimate.

[0212] In one possible embodiment, the measurement comparability matrix of the extended Kalman filter process can be as shown in Equation (16).

[0213] (16)

[0214] Here, H represents the solution method for the defined measurement comparability matrix, and the parameters included represent the parameters involved in each solution of H. The meaning of the specific parameters has been explained and will not be repeated here.

[0215] In one possible embodiment, the first measurement noise covariance matrix can be set to a preset value.

[0216] In one possible embodiment, the first measurement noise covariance matrix can be determined based on the confidence level corresponding to the tire vertical force estimate.

[0217] In this embodiment, the first measurement noise covariance matrix is ​​determined by the confidence level of the tire vertical force estimate output by the Gaussian process regression model. This first measurement noise covariance matrix can be dynamically adjusted to reduce the impact of low confidence levels in the tire vertical force estimate on the accuracy of the vehicle inertial parameter estimation, thereby improving the estimation accuracy. The combination of the Gaussian process regression model with extended Kalman filtering and subsequent Kalman filtering further enhances the overall estimation accuracy. Other methods used in related technologies to estimate vehicle inertial parameters cannot achieve the same level of dynamic adjustment of errors affecting accuracy throughout the process, resulting in lower overall accuracy compared to the method described in this embodiment.

[0218] The determination of the first measurement noise covariance matrix can be shown in the following formula (17). The solution process of formula (17) represents the defined way of solving the first measurement noise covariance matrix, without involving the parameter value of a specific time.

[0219] (17)

[0220] in, This represents the confidence level corresponding to the estimated vertical force of the left front tire. This represents the confidence level corresponding to the estimated vertical force of the right front tire. This represents the confidence level corresponding to the estimated vertical force of the right rear tire. This represents the confidence level corresponding to the estimated vertical force of the left rear tire.

[0221] 1004: Determine the observability assessment matrix based on the measurement Jacobian matrix and the first measurement noise covariance matrix.

[0222] In one possible embodiment, observability can be evaluated using the Fisher information matrix, as shown in Equation (18).

[0223] (18)

[0224] in, This represents the observability assessment matrix.

[0225] 1006: Based on the main diagonal elements of the observability evaluation matrix, evaluate the first vehicle inertial parameters for the current iteration.

[0226] 1008: If any main diagonal element is less than the first preset threshold, then pause the update process of the first vehicle inertia parameter corresponding to any main diagonal element, and use the first vehicle inertia parameter corresponding to any main diagonal element in the previous step as the first vehicle inertia parameter corresponding to any main diagonal element in the current step.

[0227] In one possible embodiment, the observability of the parameter is reflected by the main diagonal elements of F, when the... When the number of main diagonal elements is lower than the first preset threshold, the process can be paused. The parameters are updated to avoid divergence when observability is weak, which could lead to errors in the estimation of the vehicle's inertia parameters. Each parameter can be calibrated using experimental data to determine the corresponding first preset threshold.

[0228] Based on the above process, not only can the observability of the parameters be assessed, but it can also be determined whether the parameters need to be updated and whether to update them in the current iteration.

[0229] Regarding S106, Figure 11 A flowchart of an extended Kalman filtering process according to an embodiment of this disclosure is shown, such as... Figure 11 As shown, it includes the following steps:

[0230] S1102: Based on the load transfer model and the current tire vertical force estimate, construct the first measurement vector for the current iteration in the extended Kalman filter.

[0231] S1104: Obtain the parameter prediction values ​​and the first Kalman gain for the current update process in the extended Kalman filter.

[0232] S1106: Determine the predicted observation value for the current update based on the parameter prediction value of the current update process, the longitudinal acceleration and lateral acceleration of the current update process.

[0233] S1108: Based on the parameter prediction value of the current update process, the predicted observation value of the current update, the first measurement vector of the current update, and the first Kalman gain of the current update, the parameter estimate value of the current update is obtained. The parameter estimate value is the first vehicle inertial parameter.

[0234] In one possible embodiment, for the current processing step, taking step k as an example, a first measurement vector is constructed through a load transfer model and estimation is completed to obtain the first vehicle inertial parameters for the current step. During the extended Kalman filtering process, this can be based on... and The first Kalman gain is updated to determine which vehicle inertial parameter is updated in the current iteration and based on which first measurement vector. The specific process is as follows, including the prediction update process and the measurement update process.

[0235] The prediction update process for the kth time is shown in formulas (19) and (20).

[0236] (19)

[0237] (20)

[0238] in, Let represent the parameter prediction value of the k-th iteration of the Extended Kalman Filter, and let represent the parameter estimate value obtained based on the parameter estimate value of the (k-1)-th iteration of the Extended Kalman Filter. This represents the predicted error covariance value of the k-th iteration of the extended Kalman filter, and indicates that it is calculated based on the updated error covariance value of the (k-1)-th iteration of the extended Kalman filter. This represents the updated error covariance value of the extended Kalman filter at the k-th iteration; This represents the parameter estimate for the (k-1)th iteration of the extended Kalman filter. Let A represent the first process covariance matrix of the k-th iteration of the extended Kalman filter. It can be set to 0.001, and A represents the identity matrix.

[0239] The initial parameter estimates for the extended Kalman filter can be set to values ​​under vehicle preparation conditions. In this embodiment, the specific initial adoption count estimate is not limited.

[0240] The k-th measurement update process of the extended Kalman filter includes the following steps, including formulas (21) to (24).

[0241] (twenty one)

[0242] in, This represents the original Kalman gain of the k-th iteration of the extended Kalman filter. This represents the first measurement noise covariance matrix of the k-th iteration of the extended Kalman filter. Let represent the measurement Jacobian matrix of the k-th iteration of the extended Kalman filter.

[0243] (twenty two)

[0244] in, This is the symbol for the Hadamard product, which represents the element-wise multiplication of two matrices. This represents the first Kalman gain of the k-th extended Kalman filter. This indicates the first update flag bit of the k-th iteration of the extended Kalman filter. This indicates the second update flag, which is used to indicate whether the first vehicle inertial parameters participate in the current update during the determination of the first Kalman gain. This represents the second update flag bit of the k-th iteration of the extended Kalman filter, and its determination method includes: when the observability evaluation matrix of the k-th iteration contains the first update flag bit... When the number of main diagonal elements is greater than the first preset threshold The The element being 1 indicates that for the th element... The first vehicle inertial parameter is updated; otherwise, it is 0, indicating that the first... The first vehicle inertia parameter is not updated. Let U be a 3x1 matrix, where each element in the matrix is ​​1. Let W be a 1x4 matrix, where each element in the matrix is ​​1.

[0245] (twenty three)

[0246] in, Let represent the parameter estimate of the k-th iteration of the Extended Kalman Filter (EPF), and let represent the parameter prediction of the k-th iteration, the Kalman gain of the k-th iteration, the first measurement vector of the k-th iteration, and the predicted observation of the k-th iteration. This represents the predicted observation of the k-th iteration of the Extended Kalman Filter, with the input being the parameter prediction value of the k-th iteration of the Extended Kalman Filter.

[0247] (twenty four)

[0248] in, Let represent the updated error covariance value of the k-th iteration of the extended Kalman filter, and let I represent the identity matrix.

[0249] Regarding S108, Figure 12 A flowchart of another asynchronous information fusion embodiment of this disclosure is shown, such as Figure 12 As shown, it includes the following steps:

[0250] S1202: Determine the yaw acceleration based on the yaw rate in the chassis signal.

[0251] S1204: Construct a third sliding buffer based on the timestamp of yaw acceleration.

[0252] S1206: Construct a fourth sliding buffer based on the timestamp of the lateral force estimate for each tire in the tire lateral force estimate.

[0253] S1208: Construct a fifth sliding buffer based on the timestamp of the centroid longitudinal position estimate, wherein the sampling frequency of the fifth sliding buffer is the same as that of the third sliding buffer.

[0254] S1210: Obtain the first tire lateral force estimate at the current recording time in the fourth sliding buffer. The first tire lateral force estimate includes the lateral force estimate of at least one tire in the yaw dynamics model, and the number of at least one tire is less than 4.

[0255] In one possible implementation, if the first tire lateral force estimate recorded in the fourth sliding buffer at the current moment is the lateral force estimate of all four tires, then the subsequent Kalman filtering process can be executed directly.

[0256] In one possible embodiment, if the number of first tire lateral force estimates obtained is less than 4, asynchronous information fusion is required.

[0257] S1212: Get any tire lateral force estimate that has not been updated at the current recording time.

[0258] In one possible embodiment, taking the obtained lateral force estimate of the tire as FL as an example, the lateral force estimates corresponding to FR, RL, and RR have not been updated. Therefore, one of the never-updated lateral force estimates is obtained, for example, the lateral force estimate corresponding to FR.

[0259] S1214: Determine the second interpolation time based on the current recording time and the previous recording time corresponding to the lateral force estimate of each tire in the first tire lateral force estimate, as well as any tire lateral force estimate that has not been updated.

[0260] The second interpolation time is the time when there is an unupdated tire lateral force estimate in the yaw dynamics model, and it is located between the current recording time and the previous recording time. It is the recording time with the smallest time difference from the current recording time, and it is located between two recording times of other unupdated tire lateral force estimates.

[0261] The previous recording time refers to the recording time of the first tire lateral force estimate that occurred before the current recording time in the fourth sliding buffer. It should be noted that if the first tire lateral force estimate includes lateral force estimates for two tires, the previous recording time for the two tire lateral force estimates may be different. In this case, the previous recording time for the lateral force estimate of one of the tires should be used.

[0262] Among them, the other tire lateral force estimates are the tire lateral force estimates that have not been updated, except for any tire lateral force estimates that have not been updated.

[0263] In one possible embodiment, when selecting the second interpolation time in the recording time corresponding to FR, the second interpolation time needs to be between the current recording time and the previous recording time corresponding to FL, and needs to be located between the two recording times corresponding to the unupdated RL and RR.

[0264] S1216: At the second interpolation time, determine the estimated lateral force for each tire.

[0265] In one possible implementation, based on the yaw dynamics model architecture, the Kalman filtering process requires the lateral force estimates of the four tires as input, so it is necessary to determine the lateral force estimate of each tire.

[0266] S1218: Use the yaw acceleration at the second interpolation time in the third sliding buffer, the estimated lateral force of each tire at the second interpolation time in the fourth sliding buffer, and the estimated longitudinal position of the centroid at the second interpolation time in the fifth sliding buffer as the yaw acceleration, tire lateral force, and centroid longitudinal position estimates for the current time-domain alignment.

[0267] Using the above method, the time-domain aligned tire lateral force estimate is determined by interpolation, and the data at the second interpolation time in the third and fifth sliding buffers are simultaneously acquired as the current time-domain aligned yaw acceleration, tire lateral force estimate, and centroid longitudinal position estimate.

[0268] Regarding S110, Figure 13 A flowchart of a Kalman filtering process according to an embodiment of this disclosure is shown, such as... Figure 13 As shown, it includes the following steps:

[0269] S1302: Construct the second measurement vector of the Kalman filter based on the yaw dynamics model and the estimated tire lateral force.

[0270] The method for determining the second measurement vector will not be elaborated further.

[0271] S1304: Obtain the predicted parameter values ​​and the second Kalman gain for the current update process in the Kalman filter.

[0272] S1306: Construct a measurement matrix based on the current yaw acceleration.

[0273] In one possible embodiment, taking the k-th Kalman filtering process as an example, the measurement matrix for the k-th process is: .

[0274] S1308: Based on the measurement matrix, the second measurement vector, the parameter prediction value of the current update process, and the second Kalman gain of the current update, determine the parameter estimate value of the current update. The parameter estimate value is the second vehicle inertial parameter.

[0275] In one possible embodiment, the Kalman filtering process for the kth iteration includes two parts: the prediction update process for the kth iteration and the measurement update process for the kth iteration, as shown below.

[0276] The prediction update process for the kth time is as follows, including formula (25) and formula (26).

[0277] (25)

[0278] (26)

[0279] in, Let represent the parameter prediction value of the k-th iteration of the Kalman filter process, and let indicate that it is obtained based on the parameter estimate value of the (k-1)-th iteration of the Kalman filter process. This represents the parameter estimate for the (k-1)th iteration of the Kalman filter process. Let represent the predicted error covariance value of the Kalman filter at the k-th iteration, and let be the value calculated based on the updated error covariance value of the Kalman filter at the (k-1)-th iteration. This represents the updated error covariance value of the (k-1)th Kalman filter. Let A represent the covariance matrix of the second process in the k-th Kalman filter, which can be set to 0.001. Let A represent the identity matrix.

[0280] The initial parameter estimates in the Kalman filtering process can be set to values ​​from the vehicle's prepared state. This disclosure does not impose specific limitations on this aspect.

[0281] Furthermore, the measurement update process for the kth time is as follows, including formula (27) - formula (31).

[0282] (27)

[0283] in, This represents the original Kalman gain of the k-th Kalman filter. This represents the measurement matrix for the k-th Kalman filter. Let represent the second measurement noise covariance matrix of the k-th Kalman filter.

[0284] in, The second measurement noise covariance matrix represents the k-th Kalman filter. It can be set to a preset value or determined based on the confidence level corresponding to the estimated tire lateral force and the estimated longitudinal position of the centroid, as shown in formula (28).

[0285] (28)

[0286] in, This represents the confidence level corresponding to the estimated lateral force of the left front tire in the k-th iteration. This represents the confidence level corresponding to the estimated lateral force of the right front tire in the k-th iteration. This represents the confidence level corresponding to the estimated lateral force of the left rear tire. This represents the confidence level corresponding to the estimated lateral force of the right rear tire. This represents the error variance corresponding to the centroid longitudinal position estimate output in the extended Kalman filter.

[0287] (29)

[0288] in, This represents the second Kalman gain of the k-th Kalman filter. The third update flag indicates whether the second vehicle inertial parameters participate in the current update during the determination of the first Kalman gain. This represents the third update flag in the k-th iteration of the Kalman filter. Its determination method includes: when the observability evaluation value in the k-th iteration... When the number of main diagonal elements is greater than the second preset threshold The The element being 1 indicates that for the th element... The second vehicle inertial parameter is updated; otherwise, it is 0, indicating that the second vehicle inertial parameter is updated. The second vehicle inertia parameter is not updated.

[0289] (30)

[0290] in, Let represent the parameter estimate of the k-th Kalman filter, and let represent the parameter prediction of the k-th Kalman filter, the second Kalman gain of the k-th filter, the measurement matrix of the k-th filter, and the second measurement vector of the k-th filter.

[0291] (31)

[0292] in, This represents the updated error covariance value of the k-th Kalman filter, and indicates that it is calculated based on the predicted error covariance value of the k-th Kalman filter, the second Kalman gain of the k-th Kalman filter, and the measurement matrix of the k-th filter.

[0293] Among them, the observability evaluation value in formula (29) is used to evaluate the observability of the second vehicle inertial parameter. Figure 14 A flowchart illustrating the observability evaluation of a Kalman filter according to an embodiment of this disclosure is shown, such as... Figure 14 As shown, it includes the following steps:

[0294] S1402: Determine the current measurement matrix and the second measurement noise covariance matrix in the Kalman filtering process, wherein the second measurement noise covariance matrix is ​​set to a preset value or determined according to the confidence level corresponding to the tire lateral force estimate and the centroid longitudinal position estimate.

[0295] S1404: Determine the observability assessment value based on the current measurement matrix and the second measurement noise covariance matrix.

[0296] S1406: Evaluate the second vehicle inertial parameter for the current iteration based on the observability assessment value.

[0297] S1408: If the observability assessment value is less than the second preset threshold, then pause the update process of the second vehicle inertia parameter and use the previous second vehicle inertia parameter as the current second vehicle inertia parameter.

[0298] Using the above method, simulation verification was performed during the processing. The following process is the simulation verification process for the double lane change condition. The vehicle's longitudinal speed was maintained at around 50 km / h. It first traveled in a straight line, and at approximately the 5th second, it began the double lane change operation. After the double lane change was completed, it resumed traveling in a straight line.

[0299] Figure 15 A schematic diagram illustrating the yaw rate estimation effect in an embodiment of this disclosure is shown, such as... Figure 15 As shown, the table includes estimated and simulated values ​​of yaw rate and yaw acceleration, presented in three coordinate systems. The horizontal axis represents time in seconds (s) and includes values ​​of 0, 5, 10, and 15. The vertical axis of one coordinate system includes -0.2, 0, and 0.2, while the vertical axis of the other coordinate system includes -1, 0, and 1. Solid lines represent the estimated values, and dashed lines represent the simulated values. It can be seen that the estimated yaw acceleration matches the true value well, and the noise is controlled at a low level.

[0300] Figure 16 This diagram illustrates the estimation results of vehicle inertia parameters under a simulated dual lane change scenario, as shown in an embodiment of this disclosure. Figure 16 As shown, the reference and estimated values ​​for the vehicle mass, longitudinal position of the center of gravity, height of the center of gravity, and yaw moment of inertia are included. The reference values ​​are represented by solid lines, and the estimated values ​​are represented by dashed lines.

[0301] It includes 4 coordinate systems. The horizontal axis represents time, in seconds (s), with coordinates of 0, 5, 10, and 15. The vertical axes of the first coordinate system are 1000 and 1500, the vertical axes of the second coordinate system are 1 and 1.5, the vertical axes of the third coordinate system are 0.5 and 1, and the vertical axes of the fourth coordinate system are 2000 and 2500.

[0302] During the straight-line driving phase, the estimated vehicle mass and longitudinal position of the center of gravity quickly converge to near the true values. The estimated center of gravity height and yaw moment of inertia remain unchanged due to insufficient observability caused by the relatively weak vehicle dynamics, remaining at their initial values. Around the 5th second, the vehicle begins a double lane-change maneuver, increasing its dynamics. The observability assessment of the estimated center of gravity height and yaw moment of inertia reaches the required level, thus they begin to converge to near the true values.

[0303] Table 4 shows the statistical results of the simulation verification of the vehicle inertial parameter estimation. As can be seen from Table 4, the estimated values ​​of the vehicle mass and longitudinal position of the center of gravity have extremely small errors, with relative errors of 0.07% and 0.00%, respectively, indicating that the relevant parameters can be identified with high precision. The estimated value of the center of gravity height has a slightly larger error of 0.03m, with a relative error of 5.88%, but it is still within a reasonable range. The estimated value of the yaw moment of inertia is very close to the true value, with a relative error of 1.33%. Overall, the proposed estimation method can accurately identify the key inertial parameters of the vehicle in a simulation environment, verifying the effectiveness of the method.

[0304] Table 4

[0305]

[0306] In one possible embodiment, Figure 17 This diagram illustrates a tire vertical force estimation under a double lane change condition according to an embodiment of the present disclosure. It includes four coordinate systems for the vertical force estimation values ​​of the four tires: the left front tire, the right front tire, the left rear tire, and the right rear tire. Each coordinate system includes a reference value and an estimated value. The solid line represents the reference value, and the dashed line represents the estimated value. To visually demonstrate the consistency between the estimated value and the reference value in the embodiment of the present disclosure, they are no longer shown as two separate diagrams. The reference value and the estimated value are shown in one coordinate system.

[0307] The horizontal axis of each coordinate system represents time, in seconds (s), including 0, 2, 4, 6, and 8. The vertical axes represent the vertical forces on the four wheels (left front wheel, right front wheel, left rear wheel, and right rear wheel), in kN, including 5 and 10.

[0308] Figure 18 This diagram illustrates a tire lateral force estimation under a double lane change condition according to an embodiment of the present disclosure. It includes four coordinate systems for the estimated lateral force values ​​of the four tires: the left front tire, the right front tire, the left rear tire, and the right rear tire. Each coordinate system includes a reference value and an estimated value. The solid line represents the reference value, and the dashed line represents the estimated value. To visually demonstrate the consistency between the estimated value and the reference value in the embodiment of the present disclosure, they are no longer shown as two separate diagrams. The reference value and the estimated value are shown in one coordinate system.

[0309] The horizontal axis of each coordinate system represents time, in seconds (s), with values ​​of 0, 2, 4, 6, and 8. The vertical axis represents the lateral forces of the four wheels (left front wheel, right front wheel, left rear wheel, and right rear wheel), in kN, with values ​​of -5, 0, and 5.

[0310] pass Figure 17 and Figure 18 It can be seen that the tire force estimation provided by the smart tire is highly consistent with the measured value.

[0311] Figure 19 A schematic diagram illustrating the inertial parameter identification results of a real vehicle under dual lane change conditions in an embodiment of this disclosure is shown, as follows: Figure 19 As shown, it includes four coordinate systems. The horizontal axis represents time, in seconds (s), with values ​​of 0, 1, 2, 3, 4, 5, 6, 7, and 8. The vertical axis includes reference and estimated values ​​for the vehicle's mass, longitudinal position of the center of gravity, height of the center of gravity, and yaw moment of inertia. The vertical axes of the first coordinate system are 3000, 3500, 4000, and 4500; the vertical axes of the second coordinate system are 0.4, 0.6, and 0.8; the vertical axes of the third coordinate system are 1 and 1.5; and the vertical axes of the fourth coordinate system are 2000 and 2500.

[0312] It can be seen that the total mass and longitudinal position of the center of mass can quickly converge to near the true value during the initial stage of straight-line travel without longitudinal or lateral excitation. With the excitation effect of the double-line-change working condition, the height of the center of mass and the yaw moment of inertia are fully excited and eventually converge to the stable value.

[0313] In one possible embodiment, an unfamiliar dataset retained from data collected from the actual vehicle is used to verify the applicability of the method in this disclosure in real-world scenarios. The total mass and longitudinal position of the vehicle's center of gravity are measured by a vehicle weighing platform; the total mass is 2574 kg, and the distance from the center of gravity to the front axle is 1.56 m. Due to testing limitations, the true values ​​of the vehicle's center of gravity height and yaw moment of inertia cannot be directly measured. Table 5 shows the statistical results of the overall vehicle inertial parameter errors verified in the actual vehicle test.

[0314] Table 5

[0315]

[0316] Table 5 presents the statistical results of the inertial parameter errors obtained from the actual vehicle verification. Compared with the simulation, although the errors in the estimated values ​​of the vehicle mass and longitudinal position of the center of gravity increased slightly, they remained at a low level. This indicates that the method in the embodiments of this disclosure has high overall estimation accuracy.

[0317] The experimental results from the simulation and real-vehicle verifications demonstrate that the vehicle inertial parameter estimation method based on intelligent tires in this embodiment can work effectively under various operating conditions. The vehicle mass and longitudinal position of the center of gravity converge rapidly to near the true value in both uniform linear motion and double lane-change conditions, verifying the accuracy and robustness of the method under common vehicle operating states. Although the center of gravity height and yaw moment of inertia are difficult to fully identify in some conditions due to limited observability, they still exhibit good convergence characteristics in strongly excited conditions such as double lane-change. This embodiment demonstrates the applicability and effectiveness of the method in real-world driving scenarios. Furthermore, the estimation method described above can improve the accuracy of estimation and enhance the online estimation accuracy of vehicle inertial parameters without requiring sufficient external excitation and additional model parameters.

[0318] Based on the same inventive concept, this disclosure also provides a vehicle inertial parameter estimation device based on smart tires, as shown in the following embodiment. Since the principle by which this device embodiment solves the problem is similar to that of the method embodiment described above, the implementation of this device embodiment can refer to the implementation of the method embodiment described above, and repeated details will not be elaborated further.

[0319] Figure 20 This diagram illustrates the structure of a vehicle inertial parameter estimation device based on smart tires, as shown in an embodiment of this disclosure. Figure 20 As shown, the vehicle inertial parameter estimation device 200 based on smart tires includes: a first estimation unit 2001, used to process the acquired acceleration signal through the smart tire system to obtain tire vertical force estimates and tire lateral force estimates; a first alignment unit 2002, used to asynchronously fuse the longitudinal acceleration, lateral acceleration, and tire vertical force estimates in the chassis signal to obtain time-domain aligned longitudinal acceleration, lateral acceleration, and tire vertical force estimates; a second estimation unit 2003, used to input the time-domain aligned longitudinal acceleration, lateral acceleration, and tire vertical force estimates into an extended Kalman filter pre-built through a load transfer model to obtain the first vehicle inertial parameters; The vehicle inertial parameters include: estimated vehicle mass, estimated longitudinal position of center of gravity, and estimated center of gravity height; the second alignment unit 2004 is used to asynchronously fuse the estimated yaw acceleration, estimated tire lateral force, and estimated longitudinal position of center of gravity determined based on the yaw rate in the chassis signal to obtain time-domain aligned yaw acceleration, estimated tire lateral force, and estimated longitudinal position of center of gravity; the third estimation unit 2005 is used to input the time-domain aligned yaw acceleration, estimated tire lateral force, and estimated longitudinal position of center of gravity into a Kalman filter pre-built through the yaw dynamics model to obtain the second vehicle inertial parameters; the second vehicle inertial parameters include: estimated yaw moment of inertia.

[0320] Those skilled in the art will understand that various aspects of this disclosure can be implemented as a system, method, or program product. Therefore, various aspects of this disclosure can be specifically implemented in the following forms: a completely hardware implementation, a completely software implementation (including firmware, microcode, etc.), or a combination of hardware and software aspects, collectively referred to herein as a "circuit," "module," or "system."

[0321] The following reference Figure 21 To describe an electronic device 2100 according to such an embodiment of the present disclosure. Figure 21 The electronic device 2100 shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments disclosed herein.

[0322] like Figure 21 As shown, the electronic device 2100 is manifested in the form of a general-purpose computing device. The components of the electronic device 2100 may include, but are not limited to: at least one processing unit 2110, at least one storage unit 2120, and a bus 2130 connecting different system components (including storage unit 2120 and processing unit 2110).

[0323] The storage unit stores program code that can be executed by the processing unit 2110, causing the processing unit 2110 to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this disclosure. For example, the processing unit 2110 can perform the steps of any of the above-described method embodiments.

[0324] Storage unit 2120 may include readable media in the form of volatile storage units, such as random access memory (RAM) 21201 and / or cache memory 21202, and may further include read-only memory (ROM) 21203.

[0325] Storage unit 2120 may also include a program / utility 21204 having a set (at least one) program module 21205, such program module 21205 including but not limited to: operating system, one or more application programs, other program modules and program data, each or some combination of these examples may include an implementation of a network environment.

[0326] Bus 2130 can represent one or more of several types of bus structures, including a memory cell bus or memory cell controller, a peripheral bus, a graphics acceleration port, a processing unit, or a local bus using any of the various bus structures.

[0327] Electronic device 2100 can also communicate with one or more external devices 2140 (e.g., keyboard, pointing device, Bluetooth device, etc.), one or more devices that enable a user to interact with electronic device 2100, and / or any device that enables electronic device 2100 to communicate with one or more other computing devices (e.g., router, modem, etc.). This communication can be performed via input / output (I / O) interface 2150. Furthermore, electronic device 2100 can also communicate with one or more networks (e.g., local area network (LAN), wide area network (WAN), and / or public networks, such as the Internet) via network adapter 2160. As shown, network adapter 2160 communicates with other modules of electronic device 2100 via bus 2130. It should be understood that, although not shown in the figures, other hardware and / or software modules can be used in conjunction with electronic device 2100, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage systems.

[0328] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, terminal device, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0329] Specifically, according to embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as a computer program product or a computer program, which includes computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the computer device to perform the methods described above.

[0330] In exemplary embodiments of this disclosure, a computer-readable storage medium is also provided, which may be a readable signal medium or a readable storage medium. A program product capable of implementing the methods described above is stored thereon. In some possible implementations, various aspects of this disclosure may also be implemented as a program product including program code, which, when run on a terminal device, causes the terminal device to perform the steps described in the "Exemplary Methods" section of this specification according to various exemplary embodiments of this disclosure.

[0331] More specific examples of computer-readable storage media in this disclosure may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0332] In this disclosure, a computer-readable storage medium may include a data signal propagated in baseband or as part of a carrier wave, carrying readable program code. Such propagated data signals may take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of transmitting, propagating, or transmitting a program for use by or in connection with an instruction execution system, apparatus, or device.

[0333] Optionally, the program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to wireless, wired, optical fiber, RF, etc., or any suitable combination thereof.

[0334] In practical implementation, program code for performing the operations of this disclosure can be written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network, including a local area network (LAN) or a wide area network (WAN), or it can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0335] It should be noted that although several modules or units for the device used to perform actions have been mentioned in the detailed description above, this division is not mandatory. In fact, according to embodiments of this disclosure, the features and functions of two or more modules or units described above can be embodied in one module or unit. Conversely, the features and functions of one module or unit described above can be further divided and embodied by multiple modules or units.

[0336] Furthermore, although the steps of the method in this disclosure are described in a specific order in the accompanying drawings, this does not require or imply that the steps must be performed in that specific order, or that all the steps shown must be performed to achieve the desired result. Additional or alternative steps may be omitted, multiple steps may be combined into one step, and / or a step may be broken down into multiple steps.

[0337] From the above description of the embodiments, those skilled in the art will readily understand that the exemplary embodiments described herein can be implemented by software or by combining software with necessary hardware. Therefore, the technical solutions according to the embodiments of this disclosure can be embodied in the form of a software product, which can be stored in a non-volatile storage medium (such as a CD-ROM, USB flash drive, external hard drive, etc.) or on a network, including several instructions to cause a computing device (such as a personal computer, server, mobile terminal, or network device, etc.) to execute the methods according to the embodiments of this disclosure.

[0338] Other embodiments of this disclosure will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This disclosure is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this disclosure are indicated by the appended claims.

Claims

1. A method for estimating vehicle inertial parameters based on smart tires, characterized in that, The method includes: The intelligent tire system processes the acquired acceleration signals to obtain estimates of the tire vertical force and tire lateral force. The longitudinal and lateral accelerations in the chassis signal and the estimated tire vertical force are asynchronously fused to obtain time-domain aligned longitudinal accelerations, lateral accelerations and the estimated tire vertical force. The time-domain aligned longitudinal acceleration, lateral acceleration, and tire vertical force estimates are input into an extended Kalman filter pre-built using a load transfer model to obtain the first vehicle inertial parameters. The first vehicle inertial parameters include: vehicle mass estimate, center of mass longitudinal position estimate, and center of mass height estimate. The yaw acceleration determined based on the yaw rate in the chassis signal, the estimated tire lateral force, and the estimated longitudinal position of the center of gravity are asynchronously fused to obtain time-domain aligned yaw acceleration, the estimated tire lateral force, and the estimated longitudinal position of the center of gravity. The time-domain aligned yaw acceleration, the estimated tire lateral force, and the estimated longitudinal position of the center of mass are input into a Kalman filter pre-built based on the yaw dynamics model to obtain the second vehicle inertial parameters; the second vehicle inertial parameters include: the estimated yaw moment of inertia.

2. The method according to claim 1, characterized in that, The method further includes: Taking one rotation of the smart tire as a cycle, sampling is performed at preset angles to obtain acceleration signals of the smart tire for multiple cycles; the acceleration signals include: longitudinal acceleration signal, lateral acceleration signal and vertical acceleration signal.

3. The method according to claim 1, characterized in that, The process of processing the acquired acceleration signal through the intelligent tire system to obtain tire vertical force estimates and tire lateral force estimates includes: The acceleration signal is obtained by taking the first acceleration signal of each tire as an independent sample; The first acceleration signal is input into a low-pass filter to obtain the second acceleration signal; During the rotation cycle of each tire, a ground imprint area is identified, and a third acceleration signal is obtained from the second acceleration signal; Based on the partial least squares method, acceleration feature points with a correlation greater than a threshold with tire force are extracted from the third acceleration signal; The acceleration feature points are input into a trained Gaussian process regression model to obtain tire vertical force estimates and tire lateral force estimates.

4. The method according to claim 1, characterized in that, The asynchronous information fusion of longitudinal and lateral acceleration from the chassis signal and the estimated tire vertical force to obtain time-domain aligned longitudinal acceleration, lateral acceleration, and estimated tire vertical force includes: A first sliding buffer is constructed based on the timestamps of the longitudinal acceleration and the lateral acceleration; A second sliding buffer is constructed based on the timestamp of the vertical force estimate for each tire in the tire vertical force estimate; Obtain the first tire vertical force estimate at the current recording moment in the second sliding buffer; the first tire vertical force estimate includes the vertical force estimate of any one tire or the vertical force estimates of two unrelated tires in the load transfer model; Based on the current recording time and the previous recording time corresponding to the vertical force estimate of each tire in the first tire vertical force estimate, a first interpolation time is determined; the first interpolation time is the recording time with the smallest time difference from the current recording time, which is located between the current recording time and the previous recording time, and where other tire vertical force estimates exist; the other tire vertical force estimates are the tire vertical force estimates that were not sampled at the current recording time among all tire vertical force estimates, and are the tire vertical force estimates that form a measurement vector with the first tire vertical force estimate in the load transfer model; the previous recording time is the recording time of the first tire vertical force estimate before the current recording time in the second sliding buffer. Determine the estimated value of the first tire vertical force at the first interpolation moment; The estimated vertical force of the first tire at the first interpolation time in the second sliding buffer, the estimated vertical force of the other tires, the longitudinal acceleration and the lateral acceleration in the first sliding buffer are used as the estimated longitudinal acceleration, lateral acceleration and the estimated vertical force of the tire for the current time-domain alignment.

5. The method according to claim 1, characterized in that, The time-domain aligned estimates of the longitudinal acceleration, lateral acceleration, and tire vertical force are input into an extended Kalman filter pre-built using a load transfer model to obtain the first vehicle inertial parameters, including: Based on the load transfer model and the current estimated tire vertical force, the first measurement vector for the current iteration in the extended Kalman filter is constructed. Obtain the parameter prediction values ​​and the first Kalman gain of the current update process in the extended Kalman filter; Based on the parameter prediction values ​​of the current update process, the longitudinal acceleration and lateral acceleration of the current update, determine the predicted observation values ​​of the current update. Based on the parameter prediction values ​​of the current update process, the predicted observation values ​​of the current update, the first measurement vector of the current update, and the first Kalman gain of the current update, the parameter estimate of the current update is obtained; the parameter estimate is the first vehicle inertial parameter.

6. The method according to claim 5, characterized in that, The method further includes: Determine the measurement Jacobian matrix and the first measurement noise covariance matrix in the extended Kalman filtering process; wherein, the first measurement noise covariance matrix is ​​set to a preset value or determined according to the confidence level corresponding to the tire vertical force estimate; The observability assessment matrix is ​​determined based on the measurement Jacobian matrix and the first measurement noise covariance matrix. Based on the main diagonal elements of the observability evaluation matrix, evaluate the first vehicle inertial parameters for the current iteration. If any one of the main diagonal elements is less than a first preset threshold, the update process of the first vehicle inertia parameter corresponding to the first one of the main diagonal elements is paused, and the first vehicle inertia parameter corresponding to the first one of the main diagonal elements in the previous time is used as the first vehicle inertia parameter corresponding to the first one of the main diagonal elements in the current time.

7. The method according to claim 1, characterized in that, The asynchronous information fusion of the yaw acceleration determined based on the yaw rate in the chassis signal, the estimated tire lateral force, and the estimated longitudinal position of the center of gravity to obtain time-domain aligned yaw acceleration, estimated tire lateral force, and estimated longitudinal position of the center of gravity includes: The yaw rate is determined based on the yaw rate in the chassis signal; A third sliding buffer is constructed based on the timestamp of the yaw angle acceleration. A fourth sliding buffer is constructed based on the timestamp of the lateral force estimate for each tire in the tire lateral force estimate; A fifth sliding buffer is constructed based on the timestamp of the estimated longitudinal position of the centroid; wherein the sampling frequency of the fifth sliding buffer is the same as that of the third sliding buffer; Obtain the first tire lateral force estimate at the current recording moment in the fourth sliding buffer; the first tire lateral force estimate includes the lateral force estimate of at least one tire in the yaw dynamics model; the number of the at least one tire is less than 4; Obtain any tire lateral force estimate that has not been updated at the current recording time; The second interpolation time is determined based on the current recording time corresponding to the lateral force estimate of each tire in the first tire lateral force estimate, the previous recording time of the current recording time, and any tire lateral force estimate that has not been updated. The second interpolation time is the recording time with the smallest time difference between the current recording time and the previous recording time, and is located between two recording times of other tire lateral force estimates that have not been updated in the yaw dynamics model. The previous recording time is the recording time of the first tire lateral force estimate that was recorded before the current recording time in the fourth sliding buffer. The other tire lateral force estimates are tire lateral force estimates that have not been updated except for any tire lateral force estimate that has not been updated. At the second interpolation time, the estimated lateral force for each tire is determined; The yaw acceleration at the second interpolation time in the third sliding buffer, the estimated lateral force of each tire at the second interpolation time in the fourth sliding buffer, and the estimated longitudinal position of the center of mass at the second interpolation time in the fifth sliding buffer are used as the yaw acceleration, the estimated lateral force of the tire, and the estimated longitudinal position of the center of mass for the current time-domain alignment.

8. The method according to claim 1, characterized in that, The time-domain aligned yaw rate acceleration, the estimated tire lateral force, and the estimated longitudinal position of the center of gravity are input into a Kalman filter pre-built based on the yaw dynamics model to obtain the second vehicle inertial parameters, including: Based on the yaw dynamics model and the estimated tire lateral force, a second measurement vector for the Kalman filter is constructed. Obtain the predicted parameter values ​​and the second Kalman gain for the current update process in the Kalman filter; Construct a measurement matrix based on the current yaw acceleration; Based on the measurement matrix, the second measurement vector, the parameter prediction value of the current update process, and the second Kalman gain of the current update, the parameter estimate value of the current update is determined, and the parameter estimate value is the second vehicle inertial parameter.

9. The method according to claim 8, characterized in that, The method further includes: Determine the current measurement matrix and the second measurement noise covariance matrix in the Kalman filtering process; wherein the second measurement noise covariance matrix is ​​set to a preset value or determined according to the confidence level corresponding to the tire lateral force estimate and the centroid longitudinal position estimate; Based on the current measurement matrix and the second measurement noise covariance matrix, determine the observability assessment value; Based on the observability assessment value, evaluate the second vehicle inertial parameters for the current iteration; If the observability assessment value is less than the second preset threshold, the update process of the second vehicle inertia parameter is paused, and the previous second vehicle inertia parameter is used as the current second vehicle inertia parameter.

10. A vehicle inertial parameter estimation device based on smart tires, characterized in that, include: The first estimation unit is used to process the acquired acceleration signal through the intelligent tire system to obtain the estimated values ​​of the tire vertical force and the tire lateral force. The first alignment unit asynchronously fuses the longitudinal acceleration and lateral acceleration in the chassis signal and the estimated value of the tire vertical force to obtain time-domain aligned longitudinal acceleration, lateral acceleration and the estimated value of the tire vertical force. The second estimation unit is used to input the time-domain aligned longitudinal acceleration, lateral acceleration and tire vertical force estimates into an extended Kalman filter pre-built based on a load transfer model to obtain the first vehicle inertial parameters. The first vehicle inertial parameters include: estimated vehicle mass, estimated longitudinal position of center of gravity, and estimated center of gravity height; The second alignment unit is used to asynchronously fuse the yaw acceleration determined based on the yaw rate in the chassis signal, the estimated value of the tire lateral force, and the estimated value of the longitudinal position of the center of gravity to obtain time-domain aligned yaw acceleration, the estimated value of the tire lateral force, and the estimated value of the longitudinal position of the center of gravity. The third estimation unit is used to input the time-domain aligned yaw acceleration, the estimated tire lateral force, and the estimated longitudinal position of the center of mass into a Kalman filter pre-built based on the yaw dynamics model to obtain the second vehicle inertial parameters; the second vehicle inertial parameters include: the estimated yaw moment of inertia.

Images