A four-wheel drive vehicle longitudinal vehicle speed calculation method, system and vehicle

Abstract

The application discloses a four-wheel drive vehicle longitudinal vehicle speed calculation method, system and vehicle. The method comprises the following steps: acquiring multi-source sensor information; determining the confidence of each wheel based on the wheel speed information and the change rate thereof and in combination with a vehicle dynamics model; calculating the longitudinal reference speed at the center of mass corresponding to each wheel by using the kinematic conversion relationship from the wheel coordinate system to the vehicle center of mass coordinate system based on the wheel speed information, the steering wheel angle information and the yaw angular velocity information of each wheel, and weighting and fusing the longitudinal reference speed at the center of mass by using the confidence of each wheel to obtain the measured vehicle speed; calculating the model acceleration of the vehicle by using the multi-source sensor information based on the vehicle longitudinal dynamics model; establishing a Kalman filter, taking the measured vehicle speed and the model acceleration as the observation, and estimating the state of the vehicle longitudinal speed to output the calibrated vehicle longitudinal speed. The application realizes high-precision estimation of the longitudinal speed of the four-wheel drive vehicle in all working conditions.

Description

Technical Field

[0001] This invention belongs to the field of vehicle control technology, and more specifically, relates to a method, system and vehicle for calculating the longitudinal speed of a four-wheel drive vehicle. Background Technology

[0002] In the operation of four-wheel drive vehicles, accurate estimation of the vehicle's longitudinal speed is crucial for many key functions such as power distribution and stability control. Precise speed information is the foundation for the normal operation of the traction control system (TCS), electronic stability control (ESC), anti-lock braking system (ABS), and four-wheel drive torque distribution system, directly affecting the vehicle's passability, stability, and driving safety.

[0003] Currently, the most common method for acquiring vehicle speed relies on wheel speed sensors. This involves collecting wheel speed signals from all four wheels, calculating the linear velocity of each wheel based on its rolling radius, and then estimating the overall vehicle speed by averaging or selecting the minimum value. However, when four-wheel-drive vehicles face complex driving conditions, this wheel speed sensor-based speed measurement method reveals many limitations.

[0004] In off-road conditions, on slippery surfaces, or when the vehicle is being driven to its limits, wheel slippage is common. When a wheel accelerates and spins or locks up during braking, the wheel speed measured by the wheel speed sensor cannot accurately reflect the vehicle's actual speed, leading to a significant error in the vehicle speed calculated based on wheel speed. This error can severely affect the vehicle control system's accurate assessment of the vehicle's status: for example, when the four-wheel drive system distributes power, incorrect speed information may result in an unreasonable distribution of power to each wheel, reducing the vehicle's traction and stability; when the traction control system is operating, inaccurate speed data can cause the system to make incorrect intervention decisions, failing to effectively ensure vehicle safety.

[0005] Furthermore, traditional speed estimation methods often fail to adequately consider the differences in motion of the four wheels of a four-wheel drive vehicle under various operating conditions, as well as the overall dynamic characteristics of the vehicle. During operation, four-wheel drive vehicles exhibit significant differences in the actual motion of their four wheels due to factors such as varying front and rear axle load distribution, different trajectories of the inner and outer wheels during steering, and differences in the coefficient of adhesion between each wheel and the road surface. Existing speed estimation methods that simply rely on wheel speed averaging or a single wheel speed reference cannot accurately capture these differences, further reducing the accuracy of speed estimation.

[0006] Some existing technologies attempt to incorporate accelerometers for auxiliary estimation, but they typically employ only simple complementary filtering or weighted averaging, failing to fully utilize vehicle dynamics models to effectively compensate for dynamic characteristics such as wheel slippage and load transfer. In extreme conditions where wheel speed sensor signals are completely unreliable (e.g., all wheels slipping), these methods still struggle to accurately output vehicle speed.

[0007] Therefore, in order to improve the accuracy of speed estimation for four-wheel drive vehicles under various complex working conditions and meet the requirements of advanced vehicle control systems for precise speed information, it is urgent to develop a speed estimation method and system for four-wheel drive vehicles that can effectively integrate multi-source sensor information, make full use of vehicle dynamics models, and have robustness under slippage conditions. Summary of the Invention

[0008] The purpose of this invention is to propose a method, system, and vehicle for calculating the longitudinal speed of a four-wheel drive vehicle, solving the problem that existing technologies rely on a single wheel speed sensor and cannot accurately measure the longitudinal speed of the vehicle under wheel slippage conditions; and realizing high-precision estimation of the longitudinal speed of a four-wheel drive vehicle under all working conditions based on multi-source information fusion and vehicle dynamics model.

[0009] To achieve the above objectives, in a first aspect, the present invention proposes a method for calculating the longitudinal speed of a four-wheel drive vehicle, comprising: Acquire multi-source sensor information, which includes at least the wheel speed information of the four wheels, the longitudinal and lateral acceleration information at the vehicle's center of gravity, the yaw rate information, and the steering wheel angle information; Based on the wheel speed information and its rate of change of each wheel, and combined with the vehicle dynamics model, the confidence level of each wheel is determined; Based on the wheel speed information of each wheel, the steering wheel angle information, and the yaw rate information, the longitudinal reference speed at the center of gravity of each wheel is calculated using the kinematic transformation relationship from the wheel coordinate system to the vehicle center of gravity coordinate system. The longitudinal reference speed at the center of gravity is then weighted and fused using the confidence level of each wheel to obtain the measured vehicle speed. Based on the vehicle's longitudinal dynamics model, the vehicle's model acceleration is calculated using the information from the multi-source sensors. A Kalman filter is established, and the measured vehicle speed and the model acceleration are used as observations to perform state estimation of the vehicle's longitudinal speed, and the calibrated vehicle longitudinal speed is output.

[0010] Optionally, determining the confidence level of each wheel based on its wheel speed information and rate of change, combined with a vehicle dynamics model, includes: Wheel acceleration is calculated based on the wheel speed information of each wheel. The wheel acceleration is then low-pass filtered to obtain the filtered wheel acceleration. Based on the filtered wheel acceleration, a first wheel acceleration confidence score is obtained by looking up a table or mapping function. The variance between the wheel acceleration and the filtered wheel acceleration is calculated and low-pass filtered to obtain the filtered variance. Based on the filtered variance, a second wheel acceleration confidence score is obtained by looking up a table or mapping function. The minimum value between the first wheel acceleration confidence score and the second wheel acceleration confidence score is taken as the wheel acceleration confidence score. The wheel acceleration derivative is calculated based on the wheel speed information of each wheel. The wheel acceleration derivative is then low-pass filtered to obtain a filtered wheel acceleration derivative. Based on the filtered wheel acceleration derivative, a first wheel acceleration derivative confidence score is obtained through a table lookup or mapping function. The variance between the wheel acceleration derivative and the filtered wheel acceleration derivative is calculated, and this variance is low-pass filtered to obtain a filtered variance. Based on the filtered variance, a second wheel acceleration derivative confidence score is obtained through a table lookup or mapping function. The minimum value between the first and second wheel acceleration derivative confidence scores is taken as the wheel acceleration derivative confidence score. The normal force of each wheel is calculated based on the vehicle dynamics model, and the confidence level of the normal force is obtained by looking up a table or mapping function based on the normal force. The minimum value among the wheel acceleration confidence level, the wheel acceleration derivative confidence level, and the normal force confidence level is taken as the confidence level of the corresponding wheel.

[0011] Optionally, the step of calculating the longitudinal reference velocity at the center of gravity of each wheel based on the wheel speed information, the steering wheel angle information, and the yaw rate information, using the kinematic transformation relationship from the wheel coordinate system to the vehicle center of gravity coordinate system, includes: Establish a velocity transformation model between the wheel coordinate system and the vehicle center-of-mass coordinate system. For the first... For each wheel, calculate the longitudinal reference velocity at the center of mass according to the following set of equations. : When the When the left front wheel is the only wheel, the following condition is met: ; When the When the wheel is the right front wheel, the following condition is met: ; When the When the left rear wheel is the only wheel, the following condition is met: ; When the When the wheel is the right rear wheel, the following condition is met: ; in, For the first The longitudinal velocity of the vehicle's center of mass calculated from each wheel. For the first The lateral velocity of the vehicle's center of mass calculated from each wheel. , The first The longitudinal and lateral velocities of each wheel center in the wheel coordinate system. This refers to the front wheel steering angle. This is the distance from the vehicle's center of gravity to the front axle. This is the distance from the vehicle's center of gravity to the rear axle. The yaw rate is angular velocity. These correspond to the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively. The calculated longitudinal velocity of the center of mass As the first Longitudinal reference velocity at the center of mass of each wheel .

[0012] Optionally, when weighting and fusing the longitudinal reference velocity at the center of mass using the confidence level of each wheel, a weighted average formula is used: ; in, For measuring vehicle speed, For the first Confidence level of each wheel, For the first The longitudinal reference velocity at the center of mass corresponding to each wheel. These correspond to the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively.

[0013] Optionally, the calculation of the vehicle's model acceleration using the multi-source sensor information based on the established vehicle longitudinal dynamics model includes: Based on the vehicle's center of gravity velocity, yaw rate, front wheel steering angle, and vehicle geometric parameters, the lateral velocity of each wheel is obtained through kinematic transformation, and the lateral force of each wheel is calculated based on the vehicle's lateral mechanical model. Calculate the longitudinal force of each wheel based on its driving torque, braking torque, wheel moment of inertia, wheel angular acceleration, and tire rolling radius. The driving force of the entire vehicle is calculated based on the longitudinal force, lateral force, front wheel steering angle, and air resistance of each wheel. The vehicle's model acceleration is calculated based on the vehicle's driving force and mass.

[0014] Optionally, the calculation of the longitudinal force of each wheel further includes Kalman filtering estimation of the longitudinal adhesion coefficient between the tire and the road surface, specifically: Using the wheel angular velocity as the observed value, a state equation and an observation equation based on the longitudinal adhesion coefficient are constructed. The longitudinal adhesion coefficient is predicted and corrected by Kalman filtering, and the longitudinal force of the wheel is calculated using the corrected longitudinal adhesion coefficient.

[0015] Optionally, establishing a Kalman filter, using the measured vehicle speed and the model acceleration as observations, to perform state estimation of the vehicle's longitudinal speed includes: Construct a state vector, which includes the vehicle's longitudinal acceleration and longitudinal speed; Construct a state transition matrix and an observation matrix, wherein the state transition matrix is ​​determined by the sampling time and the observation matrix is ​​an identity matrix; Using the measured vehicle speed and the model acceleration as the observation inputs of the Kalman filter, the state vector is estimated iteratively through the prediction stage and the update stage, and the filtered longitudinal vehicle speed is output.

[0016] Optionally, the expression for calculating the model acceleration is: ; ; in, Accelerate the model. For the driving force of the whole vehicle, For the overall vehicle quality, These correspond to the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively. This is the front wheel steering angle (the rear wheel steering angle is zero). For the longitudinal force of the wheel, The lateral force of the wheel, This refers to air resistance.

[0017] Secondly, this invention proposes a longitudinal speed calculation system for a four-wheel drive vehicle, comprising: The information acquisition module is used to acquire information from multiple sources of sensors, including at least the wheel speed information of the four wheels, the longitudinal and lateral acceleration information at the vehicle's center of gravity, the yaw rate information, and the steering wheel angle information. The confidence calculation module is used to determine the confidence level of each wheel based on the wheel speed information and its rate of change of each wheel, combined with the vehicle dynamics model. The coordinate transformation module is used to calculate the longitudinal reference velocity at the center of mass of each wheel based on the wheel speed of each wheel, the steering wheel angle information, and the yaw rate information, using the kinematic transformation relationship from the wheel coordinate system to the vehicle center of mass coordinate system. The vehicle speed calculation module is used to perform weighted fusion of the longitudinal reference speed at the center of gravity using the confidence level of each wheel to obtain the vehicle's measured speed. The model acceleration calculation module is used to calculate the model acceleration of the vehicle based on the established vehicle longitudinal dynamics model using the information from the multi-source sensors. The Kalman filter fusion module is used to establish a Kalman filter, using the measured vehicle speed and the model acceleration as observations to perform state estimation of the vehicle's longitudinal speed and output the calibrated vehicle longitudinal speed.

[0018] Thirdly, the present invention proposes a vehicle including the longitudinal speed calculation system for a four-wheel drive vehicle described in the first aspect.

[0019] The beneficial effects of this invention are as follows: By acquiring multi-source sensor information, including wheel speed information of the four wheels, longitudinal and lateral acceleration information at the vehicle's center of gravity, yaw rate information, and steering wheel angle information, and determining the confidence level of each wheel based on its wheel speed information and rate of change combined with the vehicle dynamics model, it can effectively identify abnormal conditions such as wheel slippage and eliminate unreliable wheel speed data, thereby avoiding the problem of measurement failure of a single wheel speed sensor when slipping; by using the kinematic transformation relationship from the wheel coordinate system to the vehicle's center of gravity coordinate system to calculate the longitudinal reference velocity at the center of gravity corresponding to each wheel, and using the confidence level of each wheel to perform weighted fusion of the longitudinal reference velocity at the center of gravity, it can fully consider the different conditions of the four wheels of a four-wheel drive vehicle under different conditions. The differences in motion under various operating conditions improve the accuracy and robustness of vehicle speed measurement. By calculating the vehicle's model acceleration using multi-source sensor information based on the vehicle's longitudinal dynamics model, it can provide acceleration reference values ​​independent of wheel speed when wheel speed signals are completely unreliable, compensating for the shortcomings of vehicle speed measurement under low-adjustment or low-speed conditions. By establishing a Kalman filter to use the measured vehicle speed and model acceleration as observations for state estimation, dynamic optimal fusion of multi-source information is achieved, outputting a calibrated whole vehicle longitudinal speed. This provides high-precision and high-reliability vehicle speed signals under all operating conditions, providing accurate speed references for torque distribution control, traction control system, and drag torque control system of four-wheel drive vehicles, significantly improving vehicle passability, stability, and driving safety.

[0020] The system of the present invention has other features and advantages that will be apparent from or will be set forth in detail in the accompanying drawings and following detailed description, which together serve to explain the particular principles of the invention. Attached Figure Description

[0021] The above and other objects, features and advantages of the present invention will become more apparent from the accompanying drawings, in which like reference numerals generally denote like parts.

[0022] Figure 1 A flowchart illustrating the steps of a method for calculating the longitudinal speed of a four-wheel drive vehicle according to Embodiment 1 of the present invention is shown.

[0023] Figure 2 A flowchart illustrating the steps for calculating the confidence level of wheel acceleration according to Embodiment 1 of the present invention is shown.

[0024] Figure 3A flowchart illustrating the steps for calculating the confidence level of the derivative of wheel acceleration according to Embodiment 1 of the present invention is shown.

[0025] Figure 4 A flowchart illustrating the steps for determining the confidence level of a single tire according to Embodiment 1 of the present invention is shown.

[0026] Figure 5 A flowchart illustrating the steps for calculating and measuring vehicle speed according to Embodiment 1 of the present invention is shown.

[0027] Figure 6 A flowchart illustrating the steps for calculating the longitudinal force of a vehicle according to Embodiment 1 of the present invention is shown.

[0028] Figure 7 A flowchart of a Kalman filter according to Embodiment 1 of the present invention is shown. Detailed Implementation

[0029] The invention will now be described in more detail with reference to the accompanying drawings. While preferred embodiments of the invention are shown in the drawings, it should be understood that the invention can be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that the invention will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.

[0030] Example 1

[0031] like Figure 1 As shown, this embodiment provides a method for calculating the longitudinal speed of a four-wheel drive vehicle, including: S1. Acquire multi-source sensor information, which includes at least the wheel speed information of the four wheels, the longitudinal and lateral acceleration information at the vehicle's center of gravity, the yaw rate information, and the steering wheel angle information. Specifically, firstly, multi-source sensor information is acquired. This information includes at least the wheel speeds of the four wheels, longitudinal and lateral acceleration at the vehicle's center of gravity, yaw rate, and steering wheel angle. Wheel speed information is collected by wheel speed sensors installed at the wheel hubs to reflect the wheel's rotational speed. Longitudinal and lateral acceleration, as well as yaw rate, are collected by an inertial measurement unit (IMU) at the vehicle's center of gravity to describe the overall motion of the vehicle. Steering wheel angle information is collected by a steering wheel angle sensor to determine the steering angle of the front wheels. This multi-source sensor information describes the vehicle's driving state from different dimensions: wheel speed information directly reflects the motion of each wheel but is susceptible to slippage; IMU information reflects the vehicle's acceleration and angular velocity, is less affected by slippage but exhibits integral drift; and steering wheel angle information provides the driver's intention and wheel direction. By integrating these four types of information, a comprehensive and complementary data foundation can be provided for subsequent wheel confidence assessment, kinematic transformation, and dynamic model calculation, thereby overcoming the limitations of a single sensor under complex working conditions and laying the foundation for high-precision vehicle speed estimation.

[0032] S2. Based on the wheel speed information and its rate of change for each wheel, and in conjunction with the vehicle dynamics model, determine the confidence level for each wheel, including: The wheel acceleration is calculated based on the wheel speed information of each wheel. The wheel acceleration is then low-pass filtered to obtain the filtered wheel acceleration. The first wheel acceleration confidence score is obtained based on the filtered wheel acceleration by looking up a table or mapping function. The variance between the wheel acceleration and the filtered wheel acceleration is calculated and low-pass filtered to obtain the filtered variance. The second wheel acceleration confidence score is obtained based on the filtered variance by looking up a table or mapping function. The minimum value between the first wheel acceleration confidence score and the second wheel acceleration confidence score is taken as the wheel acceleration confidence score. The wheel acceleration derivative is calculated based on the wheel speed information of each wheel. The wheel acceleration derivative is then low-pass filtered to obtain the filtered wheel acceleration derivative. The confidence level of the first wheel acceleration derivative is obtained by looking up a table or mapping function based on the filtered wheel acceleration derivative. The variance between the wheel acceleration derivative and the filtered wheel acceleration derivative is calculated and low-pass filtered to obtain the filtered variance. The confidence level of the second wheel acceleration derivative is obtained by looking up a table or mapping function based on the filtered variance. The minimum value between the confidence level of the first wheel acceleration derivative and the confidence level of the second wheel acceleration derivative is taken as the confidence level of the wheel acceleration derivative. The normal force of each wheel is calculated based on the vehicle dynamics model, and the confidence level of the normal force is obtained by looking up a table or mapping function based on the normal force. The minimum value among the wheel acceleration confidence score, wheel acceleration derivative confidence score, and normal force confidence score is taken as the confidence score of the corresponding wheel.

[0033] Specifically, when determining the confidence level for each wheel, such as Figure 2 As shown, in this embodiment, the wheel acceleration is first calculated based on the wheel speed information of each wheel, and then low-pass filtered to eliminate high-frequency noise, resulting in the filtered wheel acceleration. The first-order low-pass filtering process, as shown in formula (1), ensures the smoothness of the wheel acceleration. The first wheel acceleration confidence level is obtained based on the filtered wheel acceleration through a preset lookup table or mapping function. This confidence level reflects the reliability of the wheel acceleration amplitude. At the same time, the difference between the wheel acceleration and the filtered wheel acceleration is calculated and squared to obtain the variance of the wheel acceleration. This variance is then low-pass filtered to obtain the filtered variance, as shown in formula (2). This filtered variance reflects the severity of the wheel acceleration fluctuation. The larger the fluctuation, the more unstable the wheel speed change and the less reliable the wheel. The second wheel acceleration confidence level is obtained based on the filtered variance through a lookup table or mapping function. Since the confidence of wheel acceleration depends on both its amplitude and the degree of fluctuation, the minimum of the confidence of the first wheel acceleration and the confidence of the second wheel acceleration is taken as the confidence of wheel acceleration, that is, the more conservative value is taken, as shown in formula (3), to ensure that the confidence is suppressed when the performance in either dimension is poor.

[0034] The filtered variance and filtered wheel acceleration characterize the reliability of wheel motion from different dimensions, therefore, their corresponding confidence levels need to be obtained through lookup tables or mapping functions. In engineering implementation, the lookup tables or mapping functions are predetermined through offline calibration or simulation analysis. The core of this approach is to map the physical state of the wheel (such as filtered wheel acceleration and filtered variance) to confidence values ​​between 0 and 1. For example, under typical working conditions (such as starting on high-friction surfaces, accelerating on low-friction surfaces, braking, and steering), signals from wheel speed sensors, IMU, and steering wheel angle are collected, while the actual vehicle speed is obtained as a reference benchmark using a high-precision GPS or optical speed sensor. By analyzing the experimental data, the correspondence between physical quantities such as wheel acceleration and variance and the actual reliability of the wheel under different working conditions is statistically determined. Alternatively, a four-wheel drive vehicle model can be built using vehicle dynamics simulation software (such as CarSim and veDYNA), different road surface adhesion coefficients (dry asphalt, wet asphalt, snow, ice) and driving conditions can be set to simulate the wheel slippage process, obtain a large amount of data such as wheel speed, acceleration, and variance under various conditions, and record whether the wheels are in a slipping state as a reference for confidence calibration.

[0035] The filtered wheel acceleration reflects the magnitude of the wheel's longitudinal acceleration. When the amplitude is too large, it usually means that the wheel is experiencing violent acceleration or deceleration, which may be a sign of skidding or brake lock-up. In this case, the confidence level should be reduced accordingly. When the amplitude is within a reasonable range, the confidence level is higher. Therefore, based on the filtered wheel acceleration, the confidence level of the first wheel acceleration is obtained through a pre-calibrated lookup table or mapping function. This mapping relationship is usually set as follows: when the absolute value of the filtered wheel acceleration is less than a first threshold, the confidence level is 1; when it is greater than a second threshold, the confidence level is 0; and when it is between the two, the confidence level decreases linearly with the increase of the amplitude.

[0036] The filtered variance reflects the severity of wheel acceleration fluctuations and is a crucial indicator for determining the stability of wheel speed signals and whether wheel slippage has occurred. When a wheel accelerates and slips or locks up on a low-traction surface, the wheel speed changes rapidly, causing drastic fluctuations in wheel acceleration, and its variance increases accordingly. Conversely, under stable driving conditions, wheel acceleration changes smoothly, and the variance is small. Based on the filtered variance, the confidence level of the second wheel acceleration is obtained through table lookup or mapping function. This mapping relationship is typically set as follows: when the filtered variance is less than a first threshold, the confidence level is 1; when it is greater than a second threshold, the confidence level is 0; and when it is between the two, the confidence level decreases linearly with increasing variance. This mapping method transforms physical quantities into intuitive confidence indicators, providing a quantitative basis for subsequent fusion with the confidence level of the first wheel acceleration.

[0037] Formula (1) is the recursive expression for a first-order low-pass filter, used to smooth wheel acceleration signals. Its specific form is as follows: ; (1) in, The input signal at the current moment is used in the wheel acceleration calculation. In the calculation of the derivative of wheel acceleration, it is ; Let be the filtered output signal at the current time, denoted as . or ; This is the filtered output value from the previous moment (i.e., the previous sampling period); These are the filter coefficients, and their values ​​range from [value range missing]. This determines the filter's response speed and smoothness. The formula is derived by taking the current input... Compared with historical output A weighted average is performed to smooth the signal, whereby... The weight is This determines the degree to which historical information influences the current output. Initial time ( )hour, It is usually set to 0 or the first input value. .

[0038] Formula (2) is the expression for the variance of wheel acceleration, used to calculate the square of the deviation between the wheel acceleration and its filtered value, specifically: ; (2) in, For the first The wheel acceleration of each wheel is determined by its wheel speed. Differentiation yields, This represents the wheel acceleration after low-pass filtering. This formula amplifies the fluctuation in wheel acceleration by calculating the square of the deviation, facilitating subsequent filtering.

[0039] Formula (3) is the fusion expression for the wheel acceleration confidence level, used to combine the confidence level of wheel acceleration amplitude and the confidence level of variance, specifically: ; (3) in, The confidence level of the first-round acceleration is obtained by looking up a table based on the filtered round acceleration. The second round acceleration confidence level is obtained by looking up a table based on the filtered variance. By taking the minimum of the two values, the confidence level is ensured to be lowered in the event of abnormal wheel acceleration amplitude or severe fluctuations, thus reliably reflecting the confidence level of the wheel condition.

[0040] like Figure 3 As shown, the wheel acceleration derivative (i.e. the rate of change of acceleration) is calculated based on the wheel speed information of each wheel. Through the same low-pass filtering (formula (1)), variance calculation and table lookup mapping process, the variance between the wheel acceleration derivative and the filtered wheel acceleration derivative is calculated and low-pass filtered to obtain the confidence level of the first wheel acceleration derivative and the confidence level of the second wheel acceleration derivative. The minimum value of the two is taken as the confidence level of the wheel acceleration derivative, as shown in formula (5). This confidence level is used to evaluate the smoothness of the wheel speed change.

[0041] Formula (4) is the variance expression for the derivative of wheel acceleration, used to calculate the square of the deviation between the derivative of wheel acceleration and its filtered value, specifically: ; (4) in, For the first The derivative of the wheel acceleration of each wheel is given by Differentiation yields, This is the derivative of wheel acceleration after low-pass filtering. This formula amplifies the fluctuation of the wheel acceleration derivative by calculating the square of the deviation, facilitating subsequent filtering and thus evaluating the smoothness of wheel speed changes.

[0042] Formula (5) is the fusion expression for the confidence level of the wheel acceleration derivative, used to combine the confidence level of the wheel acceleration derivative amplitude and the confidence level of the variance, specifically: ; (5) in, The confidence level of the first-round acceleration derivative is obtained by looking up a table based on the filtered round acceleration derivative. The confidence level of the second-round acceleration derivative is obtained by looking up a table based on the filtered variance. By taking the minimum of the two values, the confidence level is ensured to be lowered in any case where the amplitude of the wheel acceleration derivative is abnormal or fluctuates drastically, thus reliably reflecting the smoothness of changes in wheel condition.

[0043] Based on this, the normal force of each wheel is calculated using the vehicle dynamics model. Specifically, the vertical load of each wheel is calculated using parameters such as the vehicle's longitudinal acceleration, lateral acceleration, mass, wheelbase, track width, and center of gravity height, through the load transfer formulas shown in formulas (6) to (15). Formulas (6) and (7) calculate the total normal force of the front and rear axles, respectively, while formulas (8) to (11) consider the load transfer between the left and right wheels caused by lateral acceleration. Finally, the expressions for the normal force of each wheel shown in formulas (12) to (15) are derived. The confidence level of the normal force is obtained by looking up a table or mapping function based on the calculated normal force. This confidence level reflects the reliability of the load between the wheel and the road surface. When the normal force is too small, the wheel is prone to slippage, and the confidence level decreases accordingly.

[0044] Formulas (6) to (15) are used to calculate the normal force (vertical load) of each wheel of the vehicle. The specific expressions are as follows: Total normal force between front and rear axles (Formula 6-7): ; (6) ; (7) in, The total normal force on the front axle The total normal force on the rear axle Wheelbase This is the distance from the center of gravity to the front axle. This is the distance from the center of mass to the rear axle. For the overall vehicle quality, It is the acceleration due to gravity. For longitudinal acceleration, The height of the center of mass.

[0045] Considering the normal forces of each wheel due to lateral load transfer (Equation 8-11): ; (8) ; (9) ; (10) ; (11) in, These are the normal forces of the left front, right front, left rear, and right rear wheels, respectively. The front wheel track. The rear wheel track. This is lateral acceleration.

[0046] The expressions for the normal forces of each wheel (Formula 12-15): ;(12) ; (13) ;(14) ; (15) The above formula introduces longitudinal acceleration. and lateral acceleration The calculation takes into account the load transfer between the front and rear axles caused by vehicle acceleration / braking and the load transfer between the left and right wheels caused by steering, thereby accurately calculating the vertical force between each wheel and the road surface. Among them, formulas (12) to (15) are explicit expressions derived from formulas (6) to (11) simultaneously, which facilitates real-time calculation. The calculated normal force is subsequently used to determine the confidence level of the normal force and to calculate the longitudinal force and lateral force.

[0047] Finally, as Figure 4 As shown, the minimum value among the wheel acceleration confidence score, wheel acceleration derivative confidence score, and normal force confidence score is used as the confidence score of the corresponding wheel. This three-dimensional minimum-value fusion strategy can promptly reduce the confidence score of a wheel under any abnormal conditions such as sudden acceleration changes, drastic acceleration fluctuations, or insufficient normal force, thereby suppressing the interference of unreliable wheels on vehicle speed estimation in subsequent weighted fusion. This confidence score calculation method fully utilizes the dynamic characteristics of wheel speed signals and vehicle dynamics, providing a reliable quality assessment basis for multi-source information fusion.

[0048] S3. Based on the wheel speed information, steering wheel angle information and yaw rate information of each wheel, the longitudinal reference speed at the center of gravity of each wheel is calculated by using the kinematic transformation relationship from the wheel coordinate system to the vehicle center of gravity coordinate system. The longitudinal reference speed at the center of gravity is then weighted and fused using the confidence level of each wheel to obtain the measured vehicle speed. Specifically, such as Figure 5As shown, when calculating the vehicle's measured speed, the process first uses the wheel speed information, steering wheel angle information, and yaw rate information of each wheel. Then, using the kinematic transformation relationship from the wheel coordinate system to the vehicle's center of gravity coordinate system, the longitudinal reference speed at the center of gravity of each wheel is calculated. Specifically, the raw wheel speed signal provided by the wheel speed sensor is affected by wheel slippage and cannot directly reflect the vehicle's true speed. Therefore, it is necessary to combine information such as driving torque, braking torque, wheel moment of inertia, and angular acceleration to solve the true longitudinal speed of each wheel in the wheel coordinate system using dynamic equations. This process effectively eliminates the interference of slippage on wheel speed measurement. Subsequently, using the front wheel steering angle, yaw rate, and vehicle geometric parameters (such as the distance from the center of gravity to the front and rear axles, wheelbase, etc.) obtained from the steering wheel angle sensor, the true longitudinal speed of each wheel in the wheel coordinate system is converted into a longitudinal reference speed at the vehicle's center of gravity using kinematic transformation formulas. This conversion considers the differences in the motion trajectory of each wheel during steering and the influence of yaw motion on wheel speed, allowing each wheel to independently calculate an estimated longitudinal speed at its center of gravity. Since the longitudinal velocities of the center of gravity calculated from the four wheels may differ due to factors such as uneven road surface adhesion and sensor noise, a weighted fusion is required using the confidence scores of each wheel calculated in the preceding steps. Wheels with higher confidence scores have a larger weight in the fusion process, while wheels with lower confidence scores have a smaller weight. The longitudinal velocities of the center of gravity calculated from the four wheels are fused using a weighted average to obtain the final measured vehicle speed. This measured speed integrates information from all four wheels, preserving the real-time performance of the wheel speed sensors while suppressing interference from slipping wheels through the confidence score mechanism. This provides reliable observation input for subsequent Kalman filter fusion with the model acceleration.

[0049] In this step, based on the wheel speed information, steering wheel angle information, and yaw rate information of each wheel, the longitudinal reference velocity at the center of mass of each wheel is calculated using the kinematic transformation relationship from the wheel coordinate system to the vehicle center of mass coordinate system, including: Establish a velocity transformation model between the wheel coordinate system and the vehicle center-of-mass coordinate system. For the first... For each wheel, calculate the longitudinal reference velocity at the center of mass using the following system of equations. : When the When the left front wheel is the only wheel, the following condition is met: ; When the When the wheel is the right front wheel, the following condition is met: ; When the When the left rear wheel is the only wheel, the following condition is met: ; When the When the wheel is the right rear wheel, the following condition is met: ; in, For the first The longitudinal velocity of the vehicle's center of mass calculated from each wheel. For the first The lateral velocity of the vehicle's center of mass calculated from each wheel. , The first The longitudinal and lateral velocities of each wheel center in the wheel coordinate system. This refers to the front wheel steering angle. This is the distance from the vehicle's center of gravity to the front axle. This is the distance from the vehicle's center of gravity to the rear axle. The yaw rate is angular velocity. These correspond to the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively. The calculated longitudinal velocity of the center of mass As the first Longitudinal reference velocity at the center of mass of each wheel .

[0050] Specifically, in calculating the longitudinal reference velocity at the center of mass of each wheel, it is first necessary to use wheel speed sensors, drive / brake information, and vehicle dynamics models to solve for the true longitudinal velocity of each wheel in the wheel coordinate system. Specifically, the solution is obtained by simultaneously solving the following three formulas: Formula (16) defines the longitudinal slip ratio. Compared with wheel speed sensor measurement value and the actual longitudinal speed of the wheel Relationship: ; (16) Formula (17) is a linear model of the longitudinal force of the tire, establishing the longitudinal force... With slip ratio Longitudinal stiffness coefficient and wheel normal force Relationship: ; (17) Formula (18) is the wheel dynamics equation, which combines longitudinal force and driving torque. Braking torque Wheel rotational inertia Wheel angular acceleration and tire rolling radius Connecting them: ; (18) The longitudinal force on the ground is calculated from the wheel dynamics using formula (18). Substitute it into formula (17) to estimate the tire slip ratio under the current working conditions. Subsequently, based on the physical relationship of slip ratio defined by formula (16), combined with the wheel speed sensor measurements... The true longitudinal velocity of each wheel in the wheel coordinate system is calculated by algebraic transformation. (Right now This speed effectively eliminates measurement errors caused by wheel slippage. Then, combined with the estimated wheel lateral speed By utilizing the kinematic transformation relationship from the wheel coordinate system to the vehicle's center of mass coordinate system, the longitudinal reference velocity at the center of mass corresponding to each wheel is obtained. Next, the lateral velocity in the wheel coordinate system also needs to be considered. (It can be estimated through a lateral dynamics model or derived from kinematic relationships, or set to zero). Using the kinematic transformation relationship from the wheel coordinate system to the vehicle center of mass coordinate system, the longitudinal reference velocity at the center of mass corresponding to each wheel can be solved.

[0051] Establish a velocity transformation model between the wheel coordinate system and the vehicle's center-of-mass coordinate system. For the first... Each wheel, according to its position (front left, front right, rear left, rear right), uses the corresponding kinematic equations: When the When the left front wheel is the only wheel, the following condition is met: ; When the When the wheel is the right front wheel, the following condition is met: ; When the When the left rear wheel is the only wheel, the following condition is met: ; When the When the wheel is the right rear wheel, the following condition is met: ; in, , That is, the speed in the known wheel coordinate system ( The solution has been obtained using formula 16-18. (obtained through other means) These represent the longitudinal velocity and lateral velocity of the wheel center in the wheel coordinate system, respectively. This is the front wheel steering angle (the left and right wheels are approximately the same). The distance from the center of mass to the front and rear axles. ω represents the yaw rate.

[0052] The calculated longitudinal velocity of the center of mass As the first Longitudinal reference velocity at the center of mass of each wheel .

[0053] In this step, when weighting and fusing the longitudinal reference velocity at the center of mass using the confidence level of each wheel, a weighted average formula is used: ; in, For measuring vehicle speed, For the first Confidence level of each wheel, For the first The longitudinal reference velocity at the center of mass corresponding to each wheel These correspond to the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively.

[0054] Specifically, the longitudinal reference velocity at the center of mass corresponding to each wheel is obtained. Subsequently, since the calculated center-of-gravity velocities of the four wheels may differ due to uneven road surface adhesion, sensor noise, or model errors, it is necessary to use the confidence scores of each wheel obtained from the aforementioned steps. Weighted fusion is performed to obtain the final vehicle measurement speed. Specifically, a weighted average formula is used: ; (19) in, To measure the vehicle speed, For the first Confidence level of each wheel, For the first The longitudinal reference velocity at the center of mass corresponding to each wheel These correspond to the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively. The physical meaning of this formula is: wheels with higher confidence levels have a greater weight in the fusion process, while wheels with lower confidence levels have a smaller weight. This weighted averaging effectively suppresses interference from slipping or abnormally signaled wheels on vehicle speed estimation, while fully utilizing information from reliable wheels, thus obtaining robust and accurate whole-vehicle speed measurements.

[0055] S4. Based on the vehicle's longitudinal dynamics model, calculate the vehicle's model acceleration using multi-source sensor information, including: Based on the vehicle's center of gravity velocity, yaw rate, front wheel steering angle, and vehicle geometric parameters, the lateral velocity of each wheel is obtained through kinematic transformation, and the lateral force of each wheel is calculated based on the vehicle's lateral mechanical model. Calculate the longitudinal force of each wheel based on its driving torque, braking torque, wheel moment of inertia, wheel angular acceleration, and tire rolling radius. The driving force of the entire vehicle is calculated based on the longitudinal force, lateral force, front wheel steering angle, and air resistance of each wheel. The vehicle's model acceleration is calculated based on the vehicle's driving force and mass.

[0056] Specifically, based on the vehicle's longitudinal dynamics model, the model acceleration of the vehicle is calculated using information from multiple sensor sources. The specific process is as follows.

[0057] First, based on the vehicle's center of gravity velocity, yaw rate, front wheel steering angle, and vehicle geometric parameters, the lateral velocity of each wheel is obtained through kinematic transformation, and the lateral force of each wheel is calculated based on the vehicle's lateral mechanics model. The longitudinal velocity of the center of gravity output from the previous Kalman filter is then used. Lateral velocity yaw rate Combined with the front wheel steering angle and vehicle geometry parameters (front track) Rear wheel track Distance from center of gravity to front axle Distance from center of gravity to rear axle The velocity of the center of mass is converted into the longitudinal velocity of each wheel in the wheel coordinate system using formulas (20)-(27). and lateral velocity .

[0058] Formulas (20) to (23): The longitudinal velocity of each wheel in the wheel coordinate system: ; (20) ; (twenty one) ; (twenty two) ; (twenty three) in, These are the longitudinal and lateral velocities of the vehicle's center of gravity, respectively. The yaw rate is angular velocity. These are the front and rear wheel track widths, respectively. These are the distances from the center of mass to the front and rear axles, respectively. These are the steering angles of the left and right front wheels, respectively (the steering angle of the rear wheels is zero). , , , These are the longitudinal velocities of the left front wheel, right front wheel, left rear wheel, and right rear wheel in the wheel coordinate system, respectively, which are the velocity components along the rolling direction of their respective wheels. Formulas (20)-(21) reflect the projection effect of the steering angle on the longitudinal velocity of the wheels, while formulas (22)-(23) are simplified because the rear wheels do not steer.

[0059] Formulas (24) to (27): Lateral velocities of each wheel in the wheel coordinate system: ; (twenty four) (25) ; (26) ;(27) in, , , , These represent the lateral velocities of the left front wheel, right front wheel, left rear wheel, and right rear wheel in the wheel coordinate system, that is, the velocity components perpendicular to the rolling direction of the wheel (lateral).

[0060] Then, based on the definition of slip angle (Formula (28)), the lateral force of each wheel is calculated using a linear tire model (Formula (29)).

[0061] Formula (28): Side slip angle : ; (28) in These represent the front left, front right, rear left, and rear right wheels, respectively.

[0062] Formula (29): Lateral force : ; in, This refers to the lateral stiffness of the tire.

[0063] Secondly, based on the driving torque, braking torque, wheel moment of inertia, wheel angular acceleration, and tire rolling radius of each wheel, the basic longitudinal force of each wheel is calculated using formula (18). To improve the estimation accuracy of low-adhesion road surfaces, Kalman filtering is further used to estimate the longitudinal adhesion coefficient online: firstly, the effective longitudinal adhesion coefficient observation value is calculated according to formula (30). Then, dynamic coefficient correction is introduced through formula (31), and state prediction is performed through formula (32) to obtain the predicted longitudinal adhesion coefficient. Then, the wheel angular velocity is predicted using formulas (33) and (34). Then, the correction amount is calculated using formula (35), and finally, the corrected longitudinal adhesion coefficient is obtained by updating it using formula (36). The final longitudinal force is calculated using formula (37) (where the normal force is...). Determined by formulas (12)-(15).

[0064] Then, the overall vehicle driving force is calculated based on the longitudinal force, lateral force, front wheel steering angle, and air resistance of each wheel. The calculation process is as follows: Figure 6 As shown. Air resistance is calculated according to formula (38), and the driving force of the whole vehicle is synthesized according to formula (39).

[0065] Finally, based on the vehicle's driving force and mass, the model acceleration of the vehicle is calculated using formula (40). This model acceleration does not depend on the wheel speed sensor and is used as the prediction input of the Kalman filter under conditions of wheel slippage or low adhesion, for calibrating the measured vehicle speed.

[0066] Formula (30): Effective longitudinal adhesion coefficient (observed value): ; (30) in, The tire's rolling radius, To drive torque, For braking torque, For the moment of inertia of the wheel, This refers to the wheel's angular acceleration.

[0067] Formulas (31) to (36): Longitudinal adhesion coefficient Kalman filtering: ; (31) ; (32) ; (33) ; (34) ; (35) ; (36) in, This is the sampling time step, which is the time interval between two consecutive recursive calculations of the Kalman filter; Kalman gain; For the first The longitudinal adhesion coefficient of each wheel is dynamically adjusted to correct the trend of adhesion coefficient changes in the prediction step of Kalman filtering. This coefficient is typically based on the current effective adhesion coefficient. The wheel slip ratio can be obtained by looking up a table or using an empirical formula, reflecting the dynamic characteristics of road surface adhesion conditions and wheel motion. For the first Updated value of the longitudinal adhesion coefficient of each wheel; This is the updated value for the wheel angular velocity. For the first The angular velocity of each wheel is measured directly by the wheel speed sensor; For the first Predicted values ​​of the angular velocity of each wheel. For this is the first Correction amount for the longitudinal adhesion coefficient of each wheel.

[0068] Formula (37): Corrected wheel longitudinal force: ; (37) Formula (38): Air resistance: ; (38) in, This is the drag coefficient.

[0069] Formula (39) is the formula for the synthesis of the driving force of the whole vehicle. The specific expression is as follows: ; (39) in, These are the longitudinal forces on the left front wheel and the right front wheel, respectively. These are the longitudinal forces on the left and right rear wheels, respectively. These are the lateral forces of the left and right front wheels, respectively (the lateral force of the rear wheel has no longitudinal projection). These represent the steering angles of the left and right front wheels, respectively. The physical meaning of this formula is: projecting the longitudinal force of the front wheels onto the vehicle's longitudinal axis through the cosine of the steering angle, directly superimposing the longitudinal force of the rear wheels, and then subtracting air resistance and the longitudinal drag component generated by the longitudinal force of the front wheels through the sine of the steering angle, yields the net driving force on the entire vehicle. Dividing the net driving force by the vehicle's mass gives the model acceleration. This acceleration does not depend on wheel speed sensors but is recursively derived from the vehicle dynamics model, making it suitable for speed calibration under slippage or low-traction conditions.

[0070] Formula (40): Vehicle model acceleration: ; (40) in, For the overall vehicle weight.

[0071] In this step, calculating the longitudinal force of each wheel also includes estimating the longitudinal adhesion coefficient between the tire and the road surface using Kalman filtering, specifically: Using the wheel angular velocity as the observed value, a state equation and an observation equation based on the longitudinal adhesion coefficient are constructed. The longitudinal adhesion coefficient is predicted and corrected by Kalman filtering, and the longitudinal force of the wheel is calculated using the corrected longitudinal adhesion coefficient.

[0072] Specifically, when calculating the longitudinal force of each wheel, to further improve the estimation accuracy under low-adhesion road surface or wheel slippage conditions, this embodiment uses Kalman filtering to estimate the longitudinal adhesion coefficient between the tire and the road surface. Specifically, this is achieved by estimating the wheel angular velocity... For the observed values, construct a system based on the longitudinal adhesion coefficient. A discrete Kalman filter with state variables.

[0073] First, based on the wheel dynamics equations, the driving torque at the current moment is used. Braking torque Wheel angular acceleration Normal force and rolling radius The effective adhesion coefficient observation value is calculated using formula (32). , which serves as the measurement input for the filter.

[0074] In the prediction phase, dynamic coefficients are introduced using formula (33). The adhesion coefficient from the previous moment is corrected to obtain... Then, the sampling step size was used. The state is predicted in one step according to formula (34) to obtain the predicted value. Meanwhile, based on the predicted adhesion coefficient, the updated value of the wheel angular acceleration is calculated using formula (35). The predicted angular velocity is then obtained by integrating using formula (36). .

[0075] During the update phase, the measured angular velocity will be... With predicted angular velocity The deviation multiplied by the Kalman gain The correction amount for the adhesion coefficient is obtained from formula (37). Finally, the predicted value and the correction amount are added together using formula (38) to obtain the optimal estimate for the current time. After filtering, the corrected longitudinal adhesion coefficient is multiplied by the wheel normal force using formula (39) to obtain the longitudinal force of each wheel (formula 39).

[0076] Through the recursive calculation of the Kalman filter described above, the changes in road surface adhesion conditions can be tracked online, effectively suppressing wheel speed noise and model errors, thereby outputting high-precision wheel longitudinal force, providing a reliable basis for the synthesis of vehicle driving force and the calculation of model acceleration.

[0077] In this step, the expression for calculating the model acceleration is: ; ; in, Accelerate the model. For the driving force of the whole vehicle, For the overall vehicle quality, These correspond to the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively. This is the front wheel steering angle (the rear wheel steering angle is zero). For the longitudinal force of the wheel, The lateral force of the wheel, This refers to air resistance.

[0078] S5. Establish a Kalman filter, take the measured vehicle speed and model acceleration as observations, perform state estimation on the longitudinal vehicle speed, and output the calibrated longitudinal vehicle speed.

[0079] In this step, a Kalman filter is established, and the measured vehicle speed and model acceleration are used as observations to perform state estimation of the vehicle's longitudinal speed, including: Construct a state vector, which includes the vehicle's longitudinal acceleration and longitudinal speed; Construct the state transition matrix and the observation matrix. The state transition matrix is ​​determined by the sampling time, and the observation matrix is ​​an identity matrix. Using the measured vehicle speed and model acceleration as the observation inputs of the Kalman filter, the state vector is estimated iteratively through the prediction and update phases, and the filtered longitudinal vehicle speed is output.

[0080] Specifically, in order to integrate the measured vehicle speed obtained from wheel speed information The model acceleration calculated from the vehicle dynamics model The system outputs the calibrated longitudinal vehicle speed and constructs a discrete Kalman filter. This filter predicts and updates the future state based on the system's dynamic model and current measurement data.

[0081] First, the state vector and observation vector of the system are established according to the state-space equations. The state vector is shown in equation (41). The observation vector is shown in equation (42).

[0082] , , (41)

[0083] in, For state vectors, The longitudinal acceleration of the entire vehicle, i.e., the model acceleration. ; To measure vehicle speed, which is obtained by weighted fusion of wheel speed information based on confidence level. ; The derivative of the vector state. This is the derivative of the longitudinal acceleration of the entire vehicle, i.e., the rate of change of longitudinal acceleration (jerk), used to describe the dynamic characteristics of longitudinal acceleration. To measure the derivative of vehicle speed, i.e. The derivative with respect to time is used as the derivative component of the state vector in the state equation.

[0084] , (42)

[0085] in, For the observation vector, in the Kalman filter, This represents the model acceleration observation value for the current period (i.e., the longitudinal acceleration calculated from the vehicle dynamics model). ),and This represents the measured vehicle speed observation value for the current period (i.e., the overall vehicle speed obtained by weighted fusion of wheel speed information). These two quantities serve as the observation inputs to the Kalman filter, used to correct the predicted state during the update phase.

[0086] This yields the system matrix. and observation matrix The expression (formulas 44-45): (43) (44) Setting the process noise covariance matrix and measurement noise covariance matrix As shown in formulas (45) and (46): (45) (46) In the Kalman filter, , It is the process noise covariance matrix The elements, among which, The process noise variance corresponding to the acceleration state Let be the covariance between acceleration and vehicle speed. , These are calibration values, used for actual settings. and The value (usually obtained through real vehicle experiments or simulation parameter tuning). and It is the measurement noise covariance matrix The diagonal elements represent the noise variances of the model acceleration observations and the measured vehicle speed observations, respectively, and are constants determined through calibration. These parameters reflect the uncertainty of the system model and the confidence level of the sensor measurements; reasonable values ​​enable the Kalman filter to achieve optimal dynamic equilibrium in state estimation.

[0087] In the prediction phase, the acceleration output from the Kalman filter in the previous cycle is first used as the basis. and vehicle speed State prediction is performed according to formulas (47) and (48): (47) (48) in, The predicted longitudinal acceleration of the whole vehicle, i.e. the acceleration prediction value of the current cycle, is directly taken from the acceleration estimate output by the Kalman filter of the previous cycle. The sampling time step of the discrete system is the time interval between two recursive calculations of the Kalman filter, which is usually equal to the execution cycle of the vehicle control system (such as 10ms or 20ms). The predicted longitudinal vehicle speed is obtained by adding the acceleration to the speed estimate output by the Kalman filter in the previous cycle and multiplying it by the sampling step size.

[0088] Then, construct the state transition matrix: The state equation of the system can be written as: ; (49) in, For the system matrix, For the input matrix, For control input. In this scheme, since there is no external control input, it can be simplified.

[0089] Furthermore, the discretized recursive formula can be obtained from the differential relationship of the state vector: ; (50) in, The state derivative of the previous period. Let this be the sampling step size. Therefore, a state transition matrix is ​​introduced. , so that: ; (51) Combining formula (50) with the system matrix By definition, the state transition matrix can be expressed as: ; (52) in It is an identity matrix.

[0090] The system matrix given by formula (44) Substituting into formula (52), we obtain the specific form of the state transition matrix: ; (53) Subsequently, prediction and updates are performed according to the standard procedures of Kalman filtering: During the prediction phase, the prediction error covariance matrix is ​​calculated: ; (54) in, The error covariance matrix of the previous period elements, For the longitudinal acceleration of the whole vehicle The estimation error variance Covariance of acceleration and vehicle speed estimation errors ( ), The longitudinal speed of the whole vehicle The estimated error variance.

[0091] During the update phase, the Kalman gain matrix is ​​first calculated: ; (55) in, This is the gain coefficient of the longitudinal acceleration state on the model's acceleration observations, used to correct the acceleration predictions. This represents the gain coefficient of the longitudinal acceleration state on the measured vehicle speed observation. This represents the gain coefficient of the longitudinal vehicle speed state on the model's acceleration observations. This is the gain coefficient of the longitudinal vehicle speed state on the measured vehicle speed observation value; This represents the noise variance of the model's acceleration observations, i.e., the degree of confidence in the acceleration calculated by the vehicle dynamics model. To measure the noise variance of vehicle speed observations, these two values ​​are typically determined through real vehicle calibration and used to adjust the gain of the Kalman filter, thereby balancing the weights between model predictions and measurement updates.

[0092] Then, using the observations from the current period, i.e., the vehicle speed is measured. and model acceleration (The observed values ​​here are denoted as) and ), and correct the predicted state: ; (56) in, The predicted longitudinal acceleration of the vehicle in the current cycle is directly obtained from the acceleration estimate output by the Kalman filter in the previous cycle. ; The predicted longitudinal vehicle speed for the current period is obtained by adding the acceleration to the vehicle speed estimate output by the Kalman filter in the previous period and multiplying it by the sampling step size. These two predicted values ​​are compared with the observed values ​​(model acceleration and measured vehicle speed) during the update phase, and the final state estimate for the current cycle is obtained after Kalman gain correction.

[0093] Finally, update the error covariance matrix: ; (57) in, Let be the posterior error covariance matrix for the current period. For the prediction error covariance matrix The elements, among which For the variance of acceleration prediction error, The variance of the vehicle speed prediction error. Let be the covariance. This formula calculates the error covariance after Kalman gain correction, which is used for the recursion of the next cycle.

[0094] After the above prediction and update iterations, the vehicle speed component in the state vector output by the Kalman filter... This refers to the calibrated longitudinal vehicle speed. This speed combines the real-time performance of the measured speed with the robustness of the model acceleration, maintaining accuracy even under conditions of wheel slippage or low traction. The Kalman filter flowchart is as follows: Figure 7 As shown.

[0095] Example 2

[0096] This embodiment provides a system for calculating the longitudinal speed of a four-wheel drive vehicle, including: The information acquisition module is used to acquire information from multiple sources of sensors, including at least the wheel speed information of the four wheels, the longitudinal and lateral acceleration information at the vehicle's center of gravity, the yaw rate information, and the steering wheel angle information. The confidence calculation module is used to determine the confidence level of each wheel based on the wheel speed information and its rate of change of each wheel, combined with the vehicle dynamics model. The coordinate transformation module is used to calculate the longitudinal reference velocity at the center of mass of each wheel based on the wheel speed, steering wheel angle information, and yaw rate information of each wheel, using the kinematic transformation relationship from the wheel coordinate system to the vehicle center of mass coordinate system. The vehicle speed calculation module is used to perform weighted fusion of the longitudinal reference speed at the center of gravity using the confidence level of each wheel to obtain the vehicle's measured speed. The model acceleration calculation module is used to calculate the vehicle's model acceleration based on the established vehicle longitudinal dynamics model using information from multiple sensor sources. The Kalman filter fusion module is used to establish a Kalman filter, using the measured vehicle speed and model acceleration as observations to perform state estimation of the vehicle's longitudinal speed and output the calibrated vehicle longitudinal speed.

[0097] Example 3

[0098] This embodiment provides a vehicle, including a four-wheel drive vehicle longitudinal speed calculation system as described in the first aspect.

[0099] The various embodiments of the present invention have been described above. These descriptions are exemplary and not exhaustive, nor are they limited to the disclosed embodiments. Many modifications and variations will be apparent to those skilled in the art without departing from the scope and spirit of the described embodiments.

Claims

1. A method for calculating the longitudinal speed of a four-wheel drive vehicle, characterized in that, include: Acquire multi-source sensor information, which includes at least the wheel speed information of the four wheels, the longitudinal and lateral acceleration information at the vehicle's center of gravity, the yaw rate information, and the steering wheel angle information; Based on the wheel speed information and its rate of change of each wheel, and combined with the vehicle dynamics model, the confidence level of each wheel is determined; Based on the wheel speed information of each wheel, the steering wheel angle information, and the yaw rate information, the longitudinal reference speed at the center of gravity of each wheel is calculated using the kinematic transformation relationship from the wheel coordinate system to the vehicle center of gravity coordinate system. The longitudinal reference speed at the center of gravity is then weighted and fused using the confidence level of each wheel to obtain the measured vehicle speed. Based on the vehicle's longitudinal dynamics model, the vehicle's model acceleration is calculated using the information from the multi-source sensors. A Kalman filter is established, and the measured vehicle speed and the model acceleration are used as observations to perform state estimation of the vehicle's longitudinal speed, and the calibrated vehicle longitudinal speed is output.

2. The method for calculating the longitudinal speed of a four-wheel drive vehicle according to claim 1, characterized in that, The determination of the confidence level for each wheel, based on its wheel speed information and rate of change, and in conjunction with the vehicle dynamics model, includes: Wheel acceleration is calculated based on the wheel speed information of each wheel. The wheel acceleration is then low-pass filtered to obtain the filtered wheel acceleration. Based on the filtered wheel acceleration, a first wheel acceleration confidence score is obtained by looking up a table or mapping function. The variance between the wheel acceleration and the filtered wheel acceleration is calculated and low-pass filtered to obtain the filtered variance. Based on the filtered variance, a second wheel acceleration confidence score is obtained by looking up a table or mapping function. The minimum value between the first wheel acceleration confidence score and the second wheel acceleration confidence score is taken as the wheel acceleration confidence score. The wheel acceleration derivative is calculated based on the wheel speed information of each wheel. The wheel acceleration derivative is then low-pass filtered to obtain a filtered wheel acceleration derivative. Based on the filtered wheel acceleration derivative, a first wheel acceleration derivative confidence score is obtained through a table lookup or mapping function. The variance between the wheel acceleration derivative and the filtered wheel acceleration derivative is calculated, and this variance is low-pass filtered to obtain a filtered variance. Based on the filtered variance, a second wheel acceleration derivative confidence score is obtained through a table lookup or mapping function. The minimum value between the first and second wheel acceleration derivative confidence scores is taken as the wheel acceleration derivative confidence score. The normal force of each wheel is calculated based on the vehicle dynamics model, and the confidence level of the normal force is obtained by looking up a table or mapping function based on the normal force. The minimum value among the wheel acceleration confidence level, the wheel acceleration derivative confidence level, and the normal force confidence level is taken as the confidence level of the corresponding wheel.

3. The method for calculating the longitudinal speed of a four-wheel drive vehicle according to claim 1, characterized in that, Based on the wheel speed information, steering wheel angle information, and yaw rate information of each wheel, the longitudinal reference velocity at the center of gravity of each wheel is calculated using the kinematic transformation relationship from the wheel coordinate system to the vehicle's center of gravity coordinate system, including: Establish a velocity transformation model between the wheel coordinate system and the vehicle center-of-mass coordinate system. For the first... For each wheel, calculate the longitudinal reference velocity at the center of mass according to the following set of equations. : When the When the left front wheel is the only wheel, the following condition is met: ； When the When the wheel is the right front wheel, the following condition is met: ； When the When the wheel is the left rear wheel, the following condition is met: ； When the When the wheel is the right rear wheel, the following condition is met: ； in, For the first The longitudinal velocity of the vehicle's center of mass calculated from each wheel. For the first The lateral velocity of the vehicle's center of mass calculated from each wheel. , The first The longitudinal and lateral velocities of each wheel center in the wheel coordinate system. This refers to the front wheel steering angle. This is the distance from the vehicle's center of gravity to the front axle. This is the distance from the vehicle's center of gravity to the rear axle. The yaw rate is angular velocity. These correspond to the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively. The calculated longitudinal velocity of the center of mass As the first Longitudinal reference velocity at the center of mass of each wheel .

4. The method for calculating the longitudinal speed of a four-wheel drive vehicle according to claim 1, characterized in that, When weighting and fusing the longitudinal reference velocity at the center of mass using the confidence level of each wheel, a weighted average formula is used: ； in, For measuring vehicle speed, For the first Confidence level of each wheel, For the first The longitudinal reference velocity at the center of mass corresponding to each wheel. These correspond to the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively.

5. The method for calculating the longitudinal speed of a four-wheel drive vehicle according to claim 1, characterized in that, The calculation of the vehicle's model acceleration based on the established vehicle longitudinal dynamics model and the multi-source sensor information includes: Based on the vehicle's center of gravity velocity, yaw rate, front wheel steering angle, and vehicle geometric parameters, the lateral velocity of each wheel is obtained through kinematic transformation, and the lateral force of each wheel is calculated based on the vehicle's lateral mechanical model. Calculate the longitudinal force of each wheel based on its driving torque, braking torque, wheel moment of inertia, wheel angular acceleration, and tire rolling radius. The driving force of the entire vehicle is calculated based on the longitudinal force, lateral force, front wheel steering angle, and air resistance of each wheel. The vehicle's model acceleration is calculated based on the vehicle's driving force and mass.

6. The method for calculating the longitudinal speed of a four-wheel drive vehicle according to claim 5, characterized in that, The calculation of the longitudinal force for each wheel also includes Kalman filtering estimation of the longitudinal adhesion coefficient between the tire and the road surface, specifically: Using the wheel angular velocity as the observed value, a state equation and an observation equation based on the longitudinal adhesion coefficient are constructed. The longitudinal adhesion coefficient is predicted and corrected by Kalman filtering, and the longitudinal force of the wheel is calculated using the corrected longitudinal adhesion coefficient.

7. The method for calculating the longitudinal speed of a four-wheel drive vehicle according to claim 1, characterized in that, The process of establishing a Kalman filter, using the measured vehicle speed and the model acceleration as observations, and performing state estimation of the vehicle's longitudinal speed includes: Construct a state vector, which includes the vehicle's longitudinal acceleration and longitudinal speed; Construct a state transition matrix and an observation matrix, wherein the state transition matrix is ​​determined by the sampling time and the observation matrix is ​​an identity matrix; Using the measured vehicle speed and the model acceleration as the observation inputs of the Kalman filter, the state vector is estimated iteratively through the prediction stage and the update stage, and the filtered longitudinal vehicle speed is output.

8. The method for calculating the longitudinal speed of a four-wheel drive vehicle according to claim 1, characterized in that, The formula for calculating the acceleration of the model is as follows: ； ； in, Accelerate the model. For the driving force of the whole vehicle, For the overall vehicle quality, These correspond to the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively. This is the front wheel steering angle (the rear wheel steering angle is zero). For the longitudinal force of the wheel, The lateral force of the wheel, This refers to air resistance.

9. A system for calculating the longitudinal speed of a four-wheel drive vehicle, characterized in that, include: The information acquisition module is used to acquire information from multiple sources of sensors, including at least the wheel speed information of the four wheels, the longitudinal and lateral acceleration information at the vehicle's center of gravity, the yaw rate information, and the steering wheel angle information. The confidence calculation module is used to determine the confidence level of each wheel based on the wheel speed information and its rate of change of each wheel, combined with the vehicle dynamics model. The coordinate transformation module is used to calculate the longitudinal reference velocity at the center of mass of each wheel based on the wheel speed of each wheel, the steering wheel angle information, and the yaw rate information, using the kinematic transformation relationship from the wheel coordinate system to the vehicle center of mass coordinate system. The vehicle speed calculation module is used to perform weighted fusion of the longitudinal reference speed at the center of gravity using the confidence level of each wheel to obtain the vehicle's measured speed. The model acceleration calculation module is used to calculate the model acceleration of the vehicle based on the established vehicle longitudinal dynamics model using the information from the multi-source sensors. The Kalman filter fusion module is used to establish a Kalman filter, using the measured vehicle speed and the model acceleration as observations to perform state estimation of the vehicle's longitudinal speed and output the calibrated vehicle longitudinal speed.

10. A vehicle, characterized in that, Includes the longitudinal speed calculation system for four-wheel drive vehicles as described in claim 9.

Images