Vehicle load estimation method and device and storage medium

Abstract

The invention discloses a vehicle load estimation method and device and a storage medium. The method comprises the steps that first load estimation is constructed based on a vehicle longitudinal dynamics model and extended Kalman filtering; constructing a vehicle operation condition evaluation coefficient based on the vehicle operation condition influence factor, and constructing second load estimation according to the vehicle operation condition evaluation coefficient, a vehicle longitudinal dynamic model and a recursive least square method; a fused load estimate is constructed based on the vehicle dynamics parameter, the first load estimate, and the second load estimate, and a current vehicle load estimate is output with the fused load estimate and the current vehicle operating parameter. The method has the beneficial effects that the accuracy of load estimation in any running state of the vehicle is ensured, and meanwhile, the random influence of environmental noise on the recursive least square method is compensated through the vehicle running working condition influence factors, so that the accuracy of load estimation is further improved; and fusion load estimation is carried out by using the recursive least square method, so that the calculation complexity is relatively low, and the calculation power demand is reduced.

Description

Technical Field

[0001] The present application relates to the technical field of vehicle load estimation, and in particular to a vehicle load estimation method, device and storage medium. Background Art

[0002] Vehicle load estimation usually involves two methods: estimation algorithms based on vehicle models and estimation algorithms based on data-driven methods. The method based on dynamic models estimates the load by building a longitudinal dynamic model of the vehicle and combining wheel speed sensors, inertial measurement units, and vehicle inherent parameters. However, the accuracy of this method is highly dependent on the accuracy of the model and is easily affected by external factors such as road conditions and signal noise. The data-driven method uses vehicle driving data to train the model online or offline, and inputs operating parameters such as vehicle speed, acceleration, and driving force to predict the load. However, this method is limited by the size of the model training set and has high requirements on the computing power of the main control unit.

[0003] Commercial vehicle load estimation technology can accurately evaluate the load of the vehicle under different working conditions by real-time monitoring and analysis of the vehicle's operating status. Because commercial vehicles need to operate at high loads and for long periods of time, their power output is more stable than that of passenger cars. Commercial vehicles equipped with wire control brake systems can better obtain the braking pressure of each wheel drive, thereby calculating the wheel-end drive braking torque, which is more conducive to load estimation. In the load estimation process, if only rolling resistance and air resistance are considered, large estimation errors may occur. Therefore, the load estimation algorithm should comprehensively consider signals such as yaw angular velocity, lateral acceleration, steering wheel angle, etc. obtained by sensors, and adaptively adjust parameters such as the rolling resistance coefficient.

[0004] When the vehicle is on a slope, the slope resistance may interfere with the load estimation results. Therefore, during the load calculation process, it is usually necessary to estimate the slope of the vehicle at the same time. Common slope estimation methods include sensor-based methods and model-based methods. Sensor-based methods usually use acceleration sensors or angle sensors for measurement. Although they perform well in static measurements, they may not be accurate enough in dynamic measurements. Model-based methods use vehicle dynamics models and combine information such as vehicle speed and driving torque to achieve load estimation, but this method is easily affected by complex working conditions and noise.

[0005] The patent "A state prediction and estimation method for unmanned vehicles", publication number: CN112758097A, publication date: May 7, 2021, specifically discloses the use of the proposed speed prediction formula with vehicle torque correction, and the determination of the optimal parameters of the speed prediction formula through features such as acceleration and acceleration derivatives, the use of genetic algorithm neural network for training, the introduction of historical data of the vehicle's current driving conditions to timely update and optimize the acceleration and acceleration derivatives, the use of kinematic methods and dynamic methods to fuse the estimation of the road slope angle, and the use of the least squares method to estimate the vehicle mass, thereby improving the estimation accuracy of the vehicle mass and sprung mass. This solution uses genetic algorithm neural network for training, which requires continuous iterative calculations, has high computational complexity, and is relatively inefficient.

[0006] The patent "A vehicle state estimation method based on improved Kalman filtering", publication number: CN118343139A, publication date: July 16, 2024, specifically discloses that according to the vehicle dynamics characteristics of the electric drive bus, a nonlinear three-degree-of-freedom dynamic model and a tire model of longitudinal, lateral and yaw are constructed; through the Sage-Husa adaptive filtering principle, the measurement noise covariance matrix R, which is regarded as a fixed value in the traditional Kalman filter algorithm, is adjusted and corrected in real time, and the Sage-Husa coefficient is optimized according to the degree of filter divergence; based on the nonlinear three-degree-of-freedom dynamic model, an extended Kalman filter algorithm based on the Sage-Husa filtering principle is constructed to achieve vehicle driving state estimation in a noisy environment. This solution is based on the Sage-Husa filtering principle to achieve self-adaptation, but it has high requirements on the computing power of the main control unit. Summary of the invention

[0007] The present application aims to solve the technical problem that the prior art cannot take into account both the accuracy of vehicle load estimation and the computing power requirement. The present application provides a vehicle load estimation method, device and storage medium. The present application ensures the accuracy of load estimation in any operating state of the vehicle by respectively constructing a first load estimation corresponding to an extended Kalman filter and a second load estimation corresponding to a recursive least squares method, and fusing the estimated values ​​of the first load estimation and the second load estimation according to vehicle dynamic parameters. At the same time, the random influence of environmental noise on the recursive least squares method is compensated by the vehicle operating condition influencing factor, so as to further improve the accuracy of load estimation. The computational complexity of the fused load estimation using the recursive least squares method is low, the computing power requirement is reduced, and the estimation efficiency is high.

[0008] To achieve the above technical objectives, a technical solution provided by the present application is a vehicle load estimation method, comprising the following steps: constructing a first load estimation based on a vehicle longitudinal dynamics model and an extended Kalman filter; constructing a vehicle operating condition evaluation coefficient based on a vehicle operating condition influencing factor, and constructing a second load estimation based on the vehicle operating condition evaluation coefficient, a vehicle longitudinal dynamics model and a recursive least squares method; constructing a fused load estimation based on vehicle dynamic parameters, the first load estimation and the second load estimation, and outputting a current vehicle load estimation value by fusing the load estimation and the current vehicle operating parameters.

[0009] Furthermore, constructing the first load estimation based on the vehicle longitudinal dynamics model and the extended Kalman filter includes: outputting a vehicle longitudinal acceleration equation based on the vehicle longitudinal dynamics model; constructing a nonlinear state transfer function using the vehicle longitudinal acceleration equation; and constructing the first load estimation using the nonlinear state transfer function and the extended Kalman filter.

[0010] Furthermore, the construction of a nonlinear state transfer function using the vehicle longitudinal acceleration equation includes: using the vehicle longitudinal velocity, vehicle mass, and slope value as state variables in combination with the vehicle longitudinal acceleration equation to construct a state space equation; using Euler transformation to discretize the state space equation to construct a nonlinear state transfer function.

[0011] Furthermore, the construction of a vehicle operating condition evaluation coefficient based on vehicle operating condition influencing factors includes: obtaining the vehicle operating condition influencing factors based on expert experience and / or influencing feature extraction; obtaining a vehicle operating condition influencing factor proportional coefficient based on the influence of the vehicle operating condition influencing factors on the vehicle operating condition stability; and constructing a vehicle operating condition evaluation coefficient based on the vehicle operating condition influencing factors and the vehicle operating condition influencing factor proportional coefficient.

[0012] Furthermore, the method of constructing a second load estimation based on the vehicle operating condition evaluation coefficient, the vehicle longitudinal dynamics model and the recursive least squares method includes: constructing a fitting equation including a forgetting factor based on the vehicle longitudinal dynamics model and the recursive least squares method; obtaining the correlation between the vehicle operating condition evaluation coefficient and the forgetting factor based on the correlation between the forgetting factor and the vehicle operating parameters and the correlation between the vehicle operating condition evaluation coefficient and the vehicle operating parameters; and constructing the second load based on the fitting equation including the forgetting factor and the correlation between the vehicle operating condition evaluation coefficient and the forgetting factor.

[0013] Furthermore, the constructing of the fused load estimate based on the vehicle dynamic parameters, the first load estimate and the second load estimate also includes: constructing a weight relationship between the first load estimate and the second load estimate based on the correlation between the wheel slip rate, the vehicle speed and the vehicle operating condition; constructing a low-pass filter coefficient based on the correlation between the vehicle operating condition and the noise; and constructing the fused load estimate with the weight relationship between the first load estimate and the second load estimate, the first load estimate, the second load estimate and the low-pass filter coefficient.

[0014] Furthermore, it also includes: if the vehicle meets the extreme working condition, stopping the vehicle load estimation and executing the extreme working condition prompt.

[0015] Furthermore, it also includes: constructing a vehicle longitudinal dynamics model based on the vehicle physical mechanism.

[0016] Another technical solution provided by the present application is a vehicle load estimation device, which is used to implement the method as described above, including: a data acquisition unit, which is used to collect vehicle operating parameters; a storage unit, which is used to store vehicle load data; and a microcontroller, which receives the vehicle operating parameters output by the data acquisition unit and the stored vehicle load data and outputs a current vehicle load estimation value.

[0017] Another technical solution provided by the present application is a computer-readable storage medium, in which a computer program or instruction is stored. When the computer program or instruction is executed by a processing device, the above method is implemented.

[0018] The beneficial effects of the present application are as follows: In view of the problems of inaccurate vehicle dynamics model, insufficient computing power and high cost of additional sensors in vehicle load estimation in the prior art, a mass estimation method based on extended Kalman filtering and recursive least squares and its fuzzy logic fusion method are used to first build a vehicle dynamics model, and then the observed vehicle speed and slope are integrated with the model calculation results through the extended Kalman filter to solve the problem of inaccurate model; a recursive least squares estimation algorithm that is adaptive to the vehicle operating conditions is established, and the result fusion is achieved through fuzzy logic. The advantage of the lower computational complexity of RLS and fuzzy logic is used to solve the problem of insufficient computing power. At the same time, conventional vehicle-mounted sensors and controller signals are used, and there is no need to carry additional sensors or controllers, thereby improving the scope of application. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 This is a flow chart of an embodiment of a vehicle load estimation method of the present application.

[0020] Figure 2 A schematic diagram of vehicle force analysis of the vehicle load estimation method of the present application.

[0021] Figure 3Schematic diagram of the correlation between the vehicle operating condition evaluation coefficient and the forgetting factor in the vehicle load estimation method of the present application.

[0022] Figure 4 A schematic diagram of the relationship between vehicle speed and load estimation ratio in the vehicle load estimation method of the present application.

[0023] Figure 5 Schematic diagram of the relationship between slip ratio and load estimation ratio in the vehicle load estimation method of the present application.

[0024] Figure 6 A comparison diagram of the vehicle load estimation method of the present application and the actual vehicle load.

[0025] Figure 7 This is a flow chart of another embodiment of the vehicle load estimation method of the present application.

[0026] Figure 8 This is a schematic diagram of the structure of the vehicle load estimation device of the present application. DETAILED DESCRIPTION

[0027] In order to make the purpose, technical solutions and advantages of the present application more clear, the present application is further described in detail below in conjunction with the drawings and examples. It should be understood that the specific implementation method described here is only an optimal embodiment of the present application, which is only used to explain the present application and does not limit the scope of protection of the present application. All other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present application.

[0028] like Figure 1 As shown, as the first embodiment of the present application, the vehicle load estimation method includes the following steps: Constructing a first load estimation based on a vehicle longitudinal dynamics model and an extended Kalman filter; constructing a vehicle operating condition evaluation coefficient based on the vehicle operating condition influencing factor, and constructing a second load estimation according to the vehicle operating condition evaluation coefficient, a vehicle longitudinal dynamics model and a recursive least squares method; A fused load estimate is constructed based on the vehicle dynamic parameter, the first load estimate, and the second load estimate, and a current vehicle load estimate is output by fusing the load estimate and the current vehicle operating parameter.

[0029] The extended Kalman filter is suitable for complex systems with significant nonlinear dynamic characteristics, and has good estimation accuracy for nonlinear systems; the recursive least squares method is suitable for linear systems with stable dynamic characteristics that do not change with time, or systems whose parameters change slowly with time. By constructing the first load estimation corresponding to the extended Kalman filter and the second load estimation corresponding to the recursive least squares method respectively, and fusing the estimated values ​​of the first load estimation and the second load estimation according to the vehicle dynamic parameters, the accuracy of the load estimation is ensured in any operating state of the vehicle. At the same time, the random influence of environmental noise on the recursive least squares method is compensated by the vehicle operating condition influencing factor, which further improves the accuracy of the load estimation. The calculation complexity of the fused load estimation using the recursive least squares method is low, and the estimation efficiency is high.

[0030] The vehicle load estimation method further includes: The vehicle longitudinal dynamics model is constructed based on the vehicle physical mechanism.

[0031] like Figure 2 As shown in the figure, during the vehicle's driving process, the driving force F provided by the power system t Balanced with driving resistance, driving resistance consists of the following four resistances: rolling resistance F f , air resistance F w , ramp resistance F i , acceleration resistance F j , the vehicle longitudinal dynamics model is: F t =F f +F w +F i +F j ; The output torque T of the internal combustion engine or motor is output through the transmission, the main reducer, and the transmission efficiency to obtain the wheel end driving force F t : Among them, i g is the transmission ratio of the transmission, i0 is the main reducer ratio, η T is the transmission efficiency, R is the tire rolling radius, and T is the vehicle output torque.

[0032] Rolling resistance F f for: F f =Mgcosθ(Rr c +Rr v ·v x ); Where M is the vehicle mass, g is the acceleration of gravity, θ is the slope, Rr c is the rolling resistance coefficient that is independent of speed, Rr cis the speed-dependent rolling resistance coefficient, v x is the vehicle longitudinal velocity.

[0033] Air resistance F w for: Among them, C d is the air resistance coefficient, A is the frontal area, and ρ is the air density.

[0034] Slope resistance F i for: F i =Mgsinθ.

[0035] Acceleration resistance F j for: in, is the vehicle longitudinal acceleration.

[0036] According to the vehicle longitudinal dynamics model transformation, the vehicle longitudinal acceleration is:

[0037] The slope resistance calculation requires an estimate of the road slope. Usually, the vehicle is equipped with an acceleration sensor, which can provide accurate acceleration information. If the vehicle is stationary on the slope, the acceleration along the longitudinal direction of the vehicle is obtained by the acceleration sensor and the slope value is converted. When the vehicle is in a dynamic condition, the acceleration a obtained by the acceleration sensor is s The longitudinal acceleration of the vehicle The difference is the slope value θ:

[0038] In other embodiments, the slope calculation can also be achieved by building a vehicle model with three degrees of freedom, seven degrees of freedom, etc., and combining the relationship between the tire longitudinal force, lateral force, slip rate, and sideslip angle.

[0039] The vehicle acceleration can be directly obtained through the inertial measurement unit or acceleration sensor, which is a measurable state quantity. There is a nonlinear relationship between the vehicle longitudinal speed, vehicle mass, and slope value. Therefore, the extended Kalman filter (EKF) is used to estimate the vehicle mass and slope, and the state space equation is constructed with the vehicle longitudinal speed, vehicle mass, and slope value. Let x1 = v x , x3=θ, and the input variable is the driving torque u=T. Since the vehicle mass remains basically unchanged during driving and the road slope does not undergo instantaneous mutations, the state space equation is: The vehicle load and slope will not change suddenly, so their derivatives are approximately 0, while the derivative of the longitudinal vehicle speed, i.e., the longitudinal acceleration, is obtained according to the vehicle longitudinal dynamics model. That is, the first load estimation based on the vehicle longitudinal dynamics model and the extended Kalman filter includes: Output the vehicle longitudinal acceleration equation based on the vehicle longitudinal dynamics model; The nonlinear state transfer function is constructed based on the vehicle longitudinal acceleration equation; The first load estimate is constructed with a nonlinear state transfer function and an extended Kalman filter.

[0040] The nonlinear state transfer function constructed with the vehicle longitudinal acceleration equation includes: The state space equation is constructed by combining the vehicle longitudinal acceleration equation with the vehicle longitudinal velocity, vehicle mass and slope value as state variables; The state space equation is discretized using Euler transformation and the nonlinear state transfer function is constructed.

[0041] The state space equation is discretized into time intervals T using Euler transformation. s The discrete state of the state space at time k+1, that is, the nonlinear state transfer function is: Keep the observed output matrix y consistent with the nonlinear state transfer function x: Let A be the state matrix, B be the input matrix, and C be the output matrix. The state matrix A is the state function f of each nonlinear state transition i For each state variable x j The partial derivative of Then the Jacobian matrix is ​​used to process the nonlinear state transfer function to obtain: x k =Ax k-1 +Bu k-1 ; y k =Cx k-1 ; C =

[111] ; Among them, after solving the partial derivatives in the Jacobian matrix, since the states of x2 and x3 at time k are the same as those at the previous time, only the partial derivatives of themselves are 1; the partial derivatives of the longitudinal speed with respect to the states x1, x2, and x3 are: A 11 =1-T s C d Aρx 1,k-1 x2,k-1 -T s gcosx 3,k-1 R v ; A 13 =T s g((Rr c +Rr v ·x1(k-1))sin(x3(k-1))-cos(x3(k-1))); Let the directly observable state be z k ,get: z k =Hx k ; Among them, H is the observation matrix. Since the longitudinal speed and slope value of the vehicle can be directly observed through the wheel speed sensor and acceleration sensor, the observation matrix H is:

[0042] In the prediction phase at time k, the nonlinear state transfer function is used to estimate the state and covariance matrix P k - , considering the uncertainty caused by process noise Q, and calculating the Kalman gain K k : P k - =AP k-1 A T +Q; K k =P k - H T (HP k - H T +R) -1 .

[0043] In the update phase at time k, the Kalman gain K is calculated k Combined with measurement information k and predicted status So as to update the state estimate and the covariance matrix P k : P k =(IK k H k ) k - ; Here, I is the identity matrix of size 3×3.

[0044] Through the covariance matrix, Kalman gain calculation and integration of the measured value into the prior estimate according to the Kalman gain, the root state estimation and the covariance matrix are realized, and the inverse of the vehicle load can be obtained through the output matrix y at each k moment. Then the first load estimate observed by the extended Kalman filter is:

[0045] like Figure 3 As shown in FIG. 1 , the vehicle operating condition evaluation coefficients based on the vehicle operating condition influencing factors include: Obtain vehicle operating condition influencing factors based on expert experience and / or influencing feature extraction; The proportional coefficient of the vehicle operating condition influencing factor is obtained by taking the influence of the vehicle operating condition influencing factor on the stability of the vehicle operating condition; and the vehicle operating condition evaluation coefficient is constructed based on the vehicle operating condition influencing factor and the proportional coefficient of the vehicle operating condition influencing factor.

[0046] The influencing features can be extracted based on expert experience or historical vehicle operating condition data to obtain the vehicle operating condition influencing factors that affect the stability of the vehicle operating condition, and the corresponding influencing factor proportional coefficient can be obtained based on the influence, so as to obtain the vehicle operating condition evaluation coefficient to evaluate the stability of the current vehicle operating condition.

[0047] In this embodiment, the vehicle operating condition influencing factors include at least: wheel acceleration ω wi , longitudinal acceleration Lateral acceleration a y and yaw rate ω z In a four-wheel drive system, there are: n is the vehicle operating condition evaluation coefficient, k i is the proportional coefficient of the wheel acceleration change of the i-th wheel. In the four-wheel drive system, i=4. is the change in wheel acceleration of the i-th wheel, is the maximum wheel acceleration change value of the i-th wheel, k5 is the proportional coefficient of the longitudinal acceleration, is the maximum longitudinal acceleration, k6 is the proportional coefficient of the lateral acceleration, a y,max is the maximum value of the lateral acceleration, k7 is the proportional coefficient of the yaw angular velocity, ω z,max is the maximum value of the yaw angular velocity.

[0048] Since the minimum values ​​of wheel acceleration, longitudinal acceleration, lateral acceleration and yaw angular velocity are all 0, the minimum value is omitted during normalization. The normalization is performed by the ratio of the current value to the maximum value of the vehicle operating condition influencing factor, and different proportional coefficients are allocated according to the degree of influence on the vehicle operating condition, so as to obtain the stability of the current vehicle operating condition.

[0049] In other cases, the factors affecting vehicle operating conditions may also include the sideslip angle of the center of mass, the roll angle of the vehicle body, etc.

[0050] When the vehicle operating condition evaluation coefficient is smaller, it means that the vehicle acceleration is smaller, and the current vehicle operating state is relatively stable. The significance of the forgetting factor is to control the weight of the influence of new observation data and old data on parameter estimation, which can handle the characteristics of non-stationary data or system parameters changing over time. The larger the forgetting factor, the greater the influence of old data and the smaller the influence of new data. Therefore, the forgetting factor is adaptively corrected by using the vehicle operating condition evaluation coefficient, so that the adaptive change of the forgetting factor adapts to the rate of change of vehicle parameters.

[0051] The second load estimation is constructed based on the vehicle operating condition evaluation coefficient, the vehicle longitudinal dynamics model and the recursive least squares method, including: A fitting equation including a forgetting factor is constructed according to the vehicle longitudinal dynamics model and the recursive least square method; According to the correlation between the forgetting factor and the vehicle operating parameters, and the correlation between the vehicle operating condition evaluation coefficient and the vehicle operating parameters, the correlation between the vehicle operating condition evaluation coefficient and the vehicle operating parameters is obtained; A second load estimation is constructed based on a fitting equation including a forgetting factor and a correlation between a vehicle operating condition evaluation coefficient and the forgetting factor.

[0052] At time k, the longitudinal driving force of the vehicle is Let the longitudinal acceleration of the vehicle including rolling resistance and ramp resistance be Vehicle load mass θ k =M, considering air resistance as interference Then the RLS expression is: Among them, there is a strong linear relationship between the longitudinal driving force, vehicle acceleration and vehicle load. Therefore, under some working conditions, the longitudinal force and vehicle longitudinal acceleration are used to fit the vehicle mass using the recursive least squares (RLS) method according to the vehicle longitudinal dynamics model: Then, the vehicle load is fitted by RLS. First, according to the covariance matrix P at time k-1 k-1 and the longitudinal acceleration of the vehicle at time k Calculate the gain matrix K at time k k , and then use the gain matrix K k Update the vehicle load estimate M with the estimated error r,k Finally, the covariance matrix P at time k is updated according to the observation results of this period k , and get the fitting equation including the forgetting factor: Among them, K k is the gain matrix K at time k k , P k-1 is the covariance matrix at time k-1, P k is the covariance matrix at time k, M r,k is the parameter estimation matrix at time k, M r,k-1 is the parameter estimation matrix at time k-1, λ k is the forgetting factor at time k, ranging from 0 to λ k ≤1.

[0053] Then, according to the changes of the forgetting factor with the vehicle operating parameters and the changes of the vehicle operating condition evaluation coefficient with the vehicle operating parameters, the correlation between the forgetting factor and the vehicle operating condition evaluation coefficient is obtained, that is, the greater the change of the vehicle operating parameters, the more unstable the vehicle operation, the larger the forgetting factor, and the larger the vehicle operating condition evaluation coefficient. At this time, the smaller the vehicle operating condition evaluation coefficient, the greater the influence of the new observation data, and the smaller the forgetting factor. In this way, the forgetting factor is adaptively changed to compensate for the random influence of environmental noise on the recursive least squares method, further improve the adaptability of the second load estimation, and improve the accuracy of load estimation.

[0054] In this embodiment, the forgetting factor ranges from 0.9≤λ k ≤1, correspondingly there is a minimum value n1 of the vehicle operating condition evaluation coefficient and a maximum value n2 of the vehicle operating condition evaluation coefficient.

[0055] The second load estimate is constructed based on the fitting equation including the forgetting factor and the correlation between the vehicle operating condition evaluation coefficient and the forgetting factor: λ k =ζn k ; Among them, ζ is the correlation between the vehicle operating condition evaluation coefficient and the forgetting factor, n k is the vehicle operating condition evaluation coefficient at time k.

[0056] The extended Kalman filter is suitable for complex systems with significant nonlinear dynamic characteristics and has good estimation accuracy for nonlinear systems; the recursive least squares method is suitable for linear systems with stable dynamic characteristics that do not change with time, or systems with parameters that change slowly with time. RLS assumes that the model parameters are relatively stable. Using fuzzy logic based on vehicle dynamic parameters, the first load estimate M calculated by EKF is e and the second load estimate M r Assign different weights w e and w r , realizing the fusion of estimation results. At this time, constructing the fused load estimation based on the vehicle dynamic parameters, the first load estimation and the second load estimation includes: constructing the weight relationship between the first load estimation and the second load estimation based on the correlation between the wheel slip rate, the vehicle speed and the vehicle operating condition; A fused load estimate is constructed based on the weight relationship between the first load estimate and the second load estimate, the first load estimate, and the second load estimate.

[0057] like Figure 4 , Figure 5 As shown, in this embodiment, the vehicle dynamic parameters include wheel slip rate and vehicle speed. Based on the correlation between the vehicle dynamic parameters and the vehicle operating conditions, the weight relationship between the first load estimation and the second load estimation is as follows: a. When the vehicle is at a lower speed, regardless of the slip rate, the RLS algorithm converges slowly, the load estimation is inaccurate, and the first load estimation weight w is greater than the second load estimation weight. e Increase, the second load estimated weight w r reduce; b. When the vehicle is at a medium speed, the wheel slip rate is small, the tire force is in the linear range, and the vehicle is in a stable driving condition. The second load is estimated to have a specific weight w r Increase, the first load estimated weight w e reduce; c. If the vehicle is at a medium speed, and is in a rapid acceleration or deceleration condition, the tire force gradually leaves the linear region, and the vehicle deviates towards an unstable state. The first load is estimated to have a specific weight w e Increase, the second load estimated weight w r reduce; d. When the vehicle is at a higher speed, RLS is easily disturbed by external signals or road noise, load estimation, EKF weight w e Increase, w r The specific gravity decreases.

[0058] After the fuzzy logic weights based on slip rate and vehicle speed are determined, the two weights need to be normalized before calculating the vehicle load, and a first-order low-pass filter is performed with the vehicle load calculated in the previous cycle or recorded in the storage unit, where the low-pass filter coefficient w kIt is related to the stability of the whole vehicle. When the driving force or braking force of the vehicle is small, the noise of the longitudinal force and longitudinal acceleration of the whole vehicle is large, and the linear relationship is not obvious. At this time, the filter coefficient should be reduced. Therefore, the fusion load estimation based on the vehicle dynamic parameters, the first load estimation and the second load estimation also includes: Constructing a weight relationship between the first load estimate and the second load estimate based on the correlation between the wheel slip rate, the vehicle speed and the vehicle operating condition; Construct low-pass filter coefficients based on the correlation between vehicle operating conditions and noise; A fused load estimate is constructed based on the weight relationship between the first load estimate and the second load estimate, the first load estimate, the second load estimate, and a low-pass filter coefficient.

[0059] The fused load is estimated as:

[0060] Then, a current vehicle load estimation value is output based on the current vehicle operating parameters, where the current vehicle operating parameters at least include the current vehicle operating speed, wheel acceleration, longitudinal acceleration, lateral acceleration, yaw angular velocity, current vehicle driving torque, and vehicle load at the previous moment.

[0061] In other embodiments, the vehicle dynamic parameter may also be longitudinal acceleration, lateral acceleration, yaw rate, sideslip angle of center of mass, etc.

[0062] As a second embodiment of the present application, the vehicle load estimation method further includes: If the vehicle meets the extreme operating conditions, the vehicle load estimation is stopped and the extreme operating condition prompt is executed.

[0063] The extreme operating conditions include at least: the slope value exceeds the preset slope threshold; and / or, the vehicle special action is triggered; and / or, the vehicle lateral acceleration and the body roll angle exceed the preset lateral acceleration threshold and the preset roll angle threshold; and / or, the vehicle drive is less than the preset drive threshold.

[0064] When any of the extreme operating conditions is met, the vehicle is considered to have met the extreme operating conditions. At this time, the longitudinal and lateral forces of the tires may be in a nonlinear range, or the front and rear / left and right load differences of the vehicle may be too large. The vehicle load estimation is stopped and the extreme operating condition prompt is executed to remind the user of the vehicle risk, avoid vehicle load estimation occupying software calculations, and ensure the efficiency of vehicle risk calculations.

[0065] In this embodiment, the extreme operating conditions are: the vehicle slope exceeds 30%; the vehicle anti-lock braking function, traction control function, and stability control function are triggered; the vehicle lateral acceleration and body roll angle are too large; the motor driving force is too small.

[0066] like Figure 6As shown in FIG. 1 , it is a comparison diagram of the vehicle load estimation and the actual vehicle load according to the solution of the present application. Figure 7 As shown, in the present application, in order to solve the problems of inaccurate vehicle dynamics model, insufficient computing power and high cost of additional sensors in vehicle load estimation in the prior art, a mass estimation method based on extended Kalman filtering and recursive least squares and its fuzzy logic fusion method are used. First, a vehicle dynamics model is built, and the observed vehicle speed and slope are fused with the model calculation results through the extended Kalman filter to solve the problem of inaccurate model; a recursive least squares estimation algorithm that is adaptive to the vehicle operating conditions is established, and the result fusion is achieved through fuzzy logic. The advantage of the lower computational complexity of RLS and fuzzy logic is used to solve the problem of insufficient computing power. At the same time, conventional vehicle-mounted sensors and controller signals are used, and there is no need to carry additional sensors or controllers, thereby improving the scope of application.

[0067] like Figure 8 As shown, as the third embodiment of the present application, the vehicle load estimation device includes: A data acquisition unit, used for collecting vehicle operating parameters; A storage unit, used for storing vehicle load data; The microcontroller receives the vehicle operating parameters output by the data acquisition unit and the stored vehicle load data to output a current vehicle load estimation value.

[0068] When the vehicle itself is equipped with sensors, the data acquisition unit can be the sensors carried by the vehicle itself, and the sensors carried by the vehicle itself collect data related to the vehicle's operating conditions, such as pressure, vibration, tilt angle, speed, acceleration, torque, etc. These data are transmitted to the microcontroller through the communication network or CAN bus for preprocessing. The microcontroller analyzes and processes the collected data according to the preset fusion load estimation, and finally calculates the current vehicle load estimation value.

[0069] In other cases, if the vehicle itself is equipped with a cloud server, that is, when the sensor data carried by the vehicle itself is transmitted to the cloud server, the data acquisition unit is a combination of the cloud server and the sensor. The microcontroller obtains the required signals and data from the cloud server through wireless communication technology to calculate the current vehicle load estimate.

[0070] In some other embodiments, the vehicle load estimation device further includes: The signal processing unit is connected to the data acquisition unit and performs signal noise processing on the output signal of the data acquisition unit.

[0071] At this time, the data acquisition unit is connected to the signal processing unit, and the signal processing unit is connected to the microcontroller. The signal processing unit performs A / D signal processing on the sensor, and when receiving the original signal of the wheel speed sensor, performs signal noise processing, converts the original signal of the intelligent / non-intelligent wheel speed sensor into a wheel speed pulse, completes the wheel speed and wheel acceleration calculation and filtering processing, and when receiving or using the longitudinal acceleration and other signals of the inertial processing unit inside the controller, filters the acceleration signal to eliminate the original signal noise.

[0072] In some other cases, the vehicle load estimating device further includes: The communication unit is used to communicate with the vehicle controller and obtain the vehicle operating parameters in the vehicle controller.

[0073] At this time, the communication unit is connected to the microcontroller and equipped with electronic components such as CAN chips to realize CAN or CANFD message communication and obtain signals such as engine or motor driving torque, vehicle speed, etc. in on-board controllers such as the drive control unit PDCU and positioning signal GPS.

[0074] In this embodiment, the storage unit is an EEPROM, which is used to store vehicle load or other necessary values.

[0075] In the present application, the vehicle may be a commercial vehicle. Due to the fluctuations in cargo load and number of passengers, the estimation and calibration of the load of a commercial vehicle is particularly important during vehicle driving. The method and device of the present application can accurately estimate the actual load of the vehicle during driving.

[0076] As a fourth embodiment of the present application, a computer-readable storage medium is used to store a computer program or instruction. When the computer program or instruction is executed by a processing device, the above-mentioned vehicle load estimation method is implemented. The computer-readable storage medium can be any available medium that can be stored by a computing device or a data storage device such as a data center containing one or more available media. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium (e.g., a solid-state hard disk), etc.

[0077] The specific implementation methods described above are preferred implementation methods of the vehicle load estimation method, device and storage medium of the present application, and are not intended to limit the specific implementation scope of the present application. The scope of the present application includes but is not limited to the specific implementation methods. All equivalent changes made in accordance with the shape and structure of the present application are within the protection scope of the present application.

Claims

1. A vehicle load estimation method, characterized in that: The steps include: Constructing a first load estimation based on a vehicle longitudinal dynamics model and an extended Kalman filter; constructing a vehicle operating condition evaluation coefficient based on the vehicle operating condition influencing factor, and constructing a second load estimation according to the vehicle operating condition evaluation coefficient, a vehicle longitudinal dynamics model and a recursive least squares method; A fused load estimate is constructed based on the vehicle dynamic parameters, the first load estimate, and the second load estimate, and a current vehicle load estimate is output by fusing the load estimate and the current vehicle operating parameters.

2. The vehicle load estimation method according to claim 1, characterized in that: The constructing of the first load estimation based on the vehicle longitudinal dynamics model and the extended Kalman filter comprises: Output the vehicle longitudinal acceleration equation based on the vehicle longitudinal dynamics model; The nonlinear state transfer function is constructed based on the vehicle longitudinal acceleration equation; The first load estimate is constructed with a nonlinear state transfer function and an extended Kalman filter.

3. The vehicle load estimation method according to claim 2, characterized in that: The nonlinear state transfer function constructed by using the vehicle longitudinal acceleration equation includes: The state space equation is constructed by combining the vehicle longitudinal acceleration equation with the vehicle longitudinal velocity, vehicle mass and slope value as state variables; The state space equation is discretized using Euler transformation and the nonlinear state transfer function is constructed.

4. The vehicle load estimation method according to claim 1, characterized in that: The vehicle operating condition evaluation coefficient constructed based on the vehicle operating condition influencing factor includes: Obtain vehicle operating condition influencing factors based on expert experience and / or influencing feature extraction; The proportional coefficient of the vehicle operating condition influencing factor is obtained by taking the influence of the vehicle operating condition influencing factor on the stability of the vehicle operating condition; and the vehicle operating condition evaluation coefficient is constructed based on the vehicle operating condition influencing factor and the proportional coefficient of the vehicle operating condition influencing factor.

5. The vehicle load estimation method according to claim 1, characterized in that: The second load estimation is constructed according to the vehicle operating condition evaluation coefficient, the vehicle longitudinal dynamics model and the recursive least squares method, including: A fitting equation including a forgetting factor is constructed according to the vehicle longitudinal dynamics model and the recursive least square method; According to the correlation between the forgetting factor and the vehicle operating parameters, and the correlation between the vehicle operating condition evaluation coefficient and the vehicle operating parameters, the correlation between the vehicle operating condition evaluation coefficient and the vehicle operating parameters is obtained; The second load is constructed according to the fitting equation including the forgetting factor and the correlation between the vehicle operating condition evaluation coefficient and the forgetting factor.

6. The vehicle load estimation method according to claim 1, characterized in that: The constructing a fused load estimate based on the vehicle dynamic parameter, the first load estimate, and the second load estimate further includes: Constructing a weight relationship between the first load estimate and the second load estimate based on the correlation between the wheel slip rate, the vehicle speed and the vehicle operating condition; Construct low-pass filter coefficients based on the correlation between vehicle operating conditions and noise; A fused load estimate is constructed based on the weight relationship between the first load estimate and the second load estimate, the first load estimate, the second load estimate, and a low-pass filter coefficient.

7. The vehicle load estimation method according to claim 1, characterized in that: Also includes: If the vehicle meets the extreme operating conditions, the vehicle load estimation is stopped and the extreme operating condition prompt is executed.

8. The vehicle load estimation method according to claim 1, characterized in that: Also includes: The vehicle longitudinal dynamics model is constructed based on the vehicle physical mechanism.

9. A vehicle load estimation device, used to implement the method according to any one of claims 1 to 8, characterized in that: include: A data acquisition unit, used for collecting vehicle operating parameters; A storage unit, used for storing vehicle load data; The microcontroller receives the vehicle operating parameters output by the data acquisition unit and the stored vehicle load data to output a current vehicle load estimation value.

10. A computer-readable storage medium, characterized in that: The storage medium stores a computer program or instruction, and when the computer program or instruction is executed by a processing device, the method according to any one of claims 1 to 8 is implemented.

Images