A method, system, device, and storage medium for estimating road roll angle based on a disturbance observer.
By incorporating the front axle disturbance lateral force and the rate of change of the disturbance centroid sideslip angle using a disturbance observer-based method, an extended model is constructed. This solves the problems of accuracy, real-time performance, and stability in road roll angle estimation in existing technologies, achieving more accurate and real-time roll angle estimation.
Patent Information
- Application Number
- CN202611513551.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-09-28
- Publication Date
- 2026-11-17
AI Technical Summary
Existing methods for estimating road roll angles struggle to simultaneously balance estimation accuracy, real-time computation, and stability under various operating conditions. In particular, under conditions of nonlinear tire lateral force, changing road surface adhesion conditions, or dynamic steering, the estimated values are prone to fluctuations or abnormal increases.
A disturbance observer-based approach is adopted. By acquiring the lateral forces of the front and rear axles of the vehicle, dynamic equations for yaw rate and centroid sideslip angle are established. The disturbance lateral force of the front axle and the rate of change of the disturbance centroid sideslip angle are introduced to construct a disturbance extension model. The disturbance observer is used to perform state estimation and back-calculate the road roll angle.
It improves the accuracy and real-time performance of road roll angle estimation, and can characterize vehicle model errors and road roll effects in the same state estimation process, reducing computational load and numerical fluctuations at low vehicle speeds.
Smart Images

Figure CN123104663A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of vehicle state estimation and chassis electronic control technology, specifically to a method, system, device, and storage medium for estimating road roll angle based on a disturbance observer. Background Technology
[0002] When a vehicle travels on a road with a roll angle, the gravitational component caused by the road roll affects the vehicle's lateral acceleration, yaw rate, and sideslip angle. The estimated road roll angle can be used for vehicle condition monitoring and stability control; the accuracy of this estimation directly impacts the identification and control effectiveness of the vehicle's driving state.
[0003] Existing methods for estimating road roll angle mainly include integrated navigation methods, environmental perception methods, kinematic estimation methods, and dynamic estimation methods. Integrated navigation methods are costly, environmental perception methods are easily affected by weather, lighting, and road conditions, and kinematic estimation methods are sensitive to sensor noise. Dynamic estimation methods typically collect vehicle operating parameters and estimate the road roll angle using Kalman filters or observers, making them easy to integrate into the vehicle's chassis electronic control system.
[0004] However, under conditions of nonlinear tire lateral force, changing road surface adhesion, or dynamic steering, the actual vehicle response can easily deviate from the predictions of the dynamic model, causing fluctuations, drift, or abnormal increases in the estimated road roll angle. Some nonlinear estimation methods involve large computational loads, and linear observation methods have limited estimation accuracy under nonlinear conditions; speed-related calculations at low vehicle speeds can also cause numerical instability. Therefore, existing methods struggle to simultaneously balance estimation accuracy, real-time computation, and stability under multiple operating conditions. Summary of the Invention
[0005] In view of the above, the present invention aims to provide a road roll angle estimation method, system, device and storage medium based on a disturbance observer, to solve the problem that existing road roll angle estimation methods are difficult to simultaneously achieve estimation accuracy, real-time calculation and stability under multiple operating conditions.
[0006] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: In a first aspect, the present invention provides a road roll angle estimation method based on a disturbance observer, the method comprising: Obtain the lateral forces of the front axle and rear axle of the vehicle, and establish the yaw rate dynamic equation and the center of mass sideslip angle dynamic equation based on the lateral forces of the front axle and rear axle. A front axle disturbance lateral force is introduced into the front axle lateral force, and the disturbance yaw moment is determined based on the front axle disturbance lateral force. The front axle disturbance lateral force is a model variable to be estimated, used to characterize the deviation of the front axle lateral force model caused by tire nonlinearity and road condition changes.
[0007] The disturbance yaw moment is introduced into the yaw angular velocity dynamic equation and the centroid sideslip angle dynamic equation, and the rate of change of the disturbance centroid sideslip angle is introduced into the centroid sideslip angle dynamic equation to construct a disturbance extension model; The yaw rate, the sideslip angle, the disturbance yaw moment, and the rate of change of the sideslip angle are used to form an extended state vector. Based on the disturbance extension model, the extended state vector is estimated using a disturbance observer to obtain the state estimate. A preset conversion relationship is established between the rate of change of the disturbance centroid sideslip angle and the road sideslip angle. The state estimate is converted according to the preset conversion relationship to obtain the estimated road sideslip angle.
[0008] The beneficial effects of this invention are: in the disturbance extension model, the yaw rate, the centroid sideslip angle, the disturbance yaw moment and the rate of change of the disturbance centroid sideslip angle are jointly estimated, and the road roll angle is calculated back based on the rate of change of the disturbance centroid sideslip angle, so that the vehicle model error and the road roll effect can be characterized in the same state estimation process, thereby improving the accuracy and real-time performance of the road roll angle estimation.
[0009] Based on the above technical solution, the present invention can be further improved as follows.
[0010] Furthermore, the step of determining the disturbance yaw moment based on the lateral force of the front axle disturbance, and incorporating the disturbance yaw moment into the yaw rate dynamic equation and the center of mass sideslip angle dynamic equation, includes: The lateral force of the front axle disturbance is taken as the only lateral force disturbance, and the distance from the vehicle's center of gravity to the front axle is obtained. The disturbance yaw moment is obtained by performing an equivalent torque conversion based on the front axle disturbance lateral force and the distance from the vehicle's center of gravity to the front axle, and then the disturbance yaw moment is superimposed on the yaw rate dynamic equation. The equivalent lateral force is obtained by converting the disturbance yaw moment and the distance from the vehicle's center of gravity to the front axle, and then the equivalent lateral force is superimposed on the dynamic equation of the center of gravity sideslip angle.
[0011] The beneficial effects of adopting the above-mentioned further scheme are: by superimposing the lateral force disturbance amount only in the front axle lateral force, the number of disturbance state variables is controlled, and the front axle disturbance lateral force is converted into a disturbance yaw moment, providing a state basis for the estimation of vehicle model error. In addition, the disturbance yaw moment is used in the yaw motion equation, and the equivalent lateral force is used in the centroid sideslip motion equation, so that the same front axle disturbance can be characterized in two dynamic equations.
[0012] Furthermore, the step of incorporating the rate of change of the perturbation centroid sideslip angle into the centroid sideslip angle dynamic equation to construct the perturbation extended model includes: Based on the gravitational component generated by road tilt, the vehicle's longitudinal speed, and gravitational acceleration, the influence of road tilt on the vehicle's lateral motion is equivalently characterized to obtain the rate of change of the disturbance centroid sideslip angle. The rate of change of the disturbance centroid sideslip angle is superimposed on the centroid sideslip angle dynamic equation; The disturbance extension model is composed of the yaw rate dynamic equation and the center of mass sideslip angle dynamic equation.
[0013] The beneficial effect of adopting the above-mentioned further scheme is that the influence of road tilt on vehicle lateral motion is characterized by the rate of change of the disturbance centroid sideslip angle, so that the influence of road tilt can be used as a disturbance state quantity in the estimation, and state parameters are provided for the inverse calculation of road tilt angle.
[0014] Furthermore, the step of estimating the extended state vector using a perturbation observer based on the perturbation extension model to obtain state estimates includes: Based on the extended state estimate of the current control cycle, the front axle steering angle, and the longitudinal vehicle speed, the front axle lateral force and the rear axle lateral force are calculated using the Pacejka tire model. The front axle lateral force, the rear axle lateral force, the extended state estimate of the current control cycle, and the feedforward yaw moment are input into the disturbance extension model for state prediction to obtain the predicted state quantities for the next control cycle. Calculate the predicted yaw rate and predicted lateral acceleration based on the predicted state quantities; The yaw rate residual is obtained by subtracting the predicted yaw rate value from the measured yaw rate value; the lateral acceleration residual is obtained by subtracting the predicted lateral acceleration value from the measured lateral acceleration value. The predicted state variables, the yaw rate residual, and the lateral acceleration residual are input into the Romberg observer, which then outputs the state estimate for the next control cycle.
[0015] The advantages of adopting the above-mentioned further scheme are: the front axle lateral force and the rear axle lateral force are calculated by the Pacejka tire model, the state variables are predicted by the perturbation extended model, and the predicted state variables are corrected by the residual between the measured values and the predicted values, so that the state estimation process can reflect the nonlinear changes of the tire lateral force, while reducing the computational load of online state estimation.
[0016] Furthermore, before the state estimate for the next control cycle is output by the Romberg observer, the following steps are also included: The perturbation extended model is subjected to Jacobian linearization to obtain the system matrix and output matrix; Observability analysis is performed based on the system matrix and the output matrix; when the observability analysis results show that the linearized perturbation extension model satisfies the observability condition, a coordinate transformation matrix is constructed. The system matrix and the output matrix are transformed using the coordinate transformation matrix to form a canonical observer model. By combining the observer canonical model and the observer's desired poles, a canonical gain matrix is obtained; The observer gain matrix is generated by performing an inverse coordinate transformation on the normalized gain matrix using the coordinate transformation matrix.
[0017] The advantages of adopting the above-mentioned further scheme are: it verifies whether the state variables in the perturbation extended model can be reconstructed through observability analysis, and obtains the observer gain matrix through observer canonical form transformation, pole placement and inverse coordinate transformation, thereby reducing the difficulty of solving the observer gain matrix and ensuring the convergence of the state estimation process.
[0018] Furthermore, after obtaining the observer gain matrix, the following is also included: The desired lateral acceleration and desired yaw rate are determined by using a single-track model. The dynamic driving coefficient is determined based on the deviation between the measured lateral acceleration and the desired lateral acceleration, and the deviation between the measured yaw rate and the desired yaw rate. The predicted value of the rate of change of the disturbance centroid sideslip angle in the current control cycle is corrected by using the preset observer correction gain, and a preliminary estimate is obtained; When the preliminary estimate is greater than the state estimate of the disturbance centroid sideslip angle change rate in the previous control cycle, the gain element used to correct the disturbance centroid sideslip angle change rate in the preset observer correction gain is scaled according to the dynamic driving coefficient, and the predicted value is corrected using the scaled gain element to obtain the state estimate of the disturbance centroid sideslip angle change rate in the current control cycle. When the preliminary estimate is less than or equal to the state estimate of the rate of change of the disturbance centroid sideslip angle in the previous control cycle, the preliminary estimate is used as the state estimate of the rate of change of the disturbance centroid sideslip angle in the current control cycle.
[0019] The beneficial effects of adopting the above-mentioned further scheme are: generating dynamic driving coefficients based on the deviation between the actual vehicle response and the expected response of the single-track model, and adjusting the observer correction gain according to the state estimation results of adjacent control cycles, reducing the interference of dynamic steering response on the road roll angle estimation results, and enabling the road roll angle estimation value to fall back normally.
[0020] Furthermore, the estimation of the extended state vector based on the perturbation extension model using a perturbation observer includes: Compare the vehicle's longitudinal speed with a preset speed threshold; When the longitudinal speed of the vehicle is lower than the preset speed threshold, the kinematic yaw rate and the kinematic center of mass sideslip angle are calculated according to the front axle steering angle and the longitudinal speed of the vehicle, based on the Ackermann kinematics relationship. The low-speed extended state vector is composed of the kinematic yaw rate, the kinematic center of mass sideslip angle, the disturbance yaw moment assigned to zero, and the disturbance center of mass sideslip angle change rate assigned to zero. The low-speed extended state vector is replaced with the extended state vector estimated based on the disturbance extension model to obtain the low-speed state estimate.
[0021] The beneficial effects of adopting the above-mentioned further scheme are: using Ackerman kinematics to obtain the vehicle motion state at low vehicle speeds and assigning zero to the disturbance state quantity helps to reduce numerical fluctuations at low vehicle speeds and improve the stability of the road roll angle estimation process.
[0022] Secondly, the present invention provides a road roll angle estimation system based on a disturbance observer, the system comprising: The basic dynamics model establishment module is used to obtain the front axle lateral force and the rear axle lateral force of the vehicle, and to establish the yaw rate dynamic equation and the center of mass sideslip angle dynamic equation based on the front axle lateral force and the rear axle lateral force. The disturbance yaw moment establishment module is used to introduce the front axle disturbance lateral force into the front axle lateral force, determine the disturbance yaw moment based on the front axle disturbance lateral force, and introduce the disturbance yaw moment into the yaw angular velocity dynamic equation and the center of mass sideslip angle dynamic equation. The perturbation extension model building module is used to introduce the rate of change of the perturbation centroid sideslip angle into the centroid sideslip angle dynamic equation to construct the perturbation extension model; An extended state estimation module is used to form an extended state vector from the yaw rate, the centroid sideslip angle, the disturbance yaw moment, and the rate of change of the disturbance centroid sideslip angle. Based on the disturbance extended model, the extended state vector is estimated using a disturbance observer to obtain a state estimate. The road tilt angle calculation module is used to establish a preset conversion relationship between the rate of change of the disturbance centroid sideslip angle and the road tilt angle, and to convert the state estimate value according to the preset conversion relationship to obtain the estimated value of the road tilt angle.
[0023] Thirdly, embodiments of the present invention provide a computer device, including a memory and a processor, wherein the memory stores a computer program, and the computer program, when executed by the processor, implements any step of the road roll angle estimation method based on a disturbance observer as described in the first aspect of the present invention.
[0024] Fourthly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the road roll angle estimation method based on a disturbance observer as described in the first aspect of the present invention. Attached Figure Description
[0025] Figure 1 This is a schematic diagram of the road roll angle estimation method provided in an embodiment of the present invention; Figure 2 This is a block diagram of a road roll angle estimation system provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of vehicle dynamics and disturbance settings provided in an embodiment of the present invention; Figure 4 The diagram shows the prediction and correction structure of the disturbance observer provided in the embodiment of the present invention. Detailed Implementation
[0026] The principles and features of the present invention are described below. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.
[0027] Example 1 This embodiment relates to a road roll angle estimation method based on a disturbance observer. For example... Figure 1 As shown, the method includes the following procedures.
[0028] Obtain the lateral forces of the front axle and rear axle of the vehicle, and establish the yaw rate dynamic equation and the center of mass sideslip angle dynamic equation of the vehicle based on the lateral forces of the front axle and rear axle.
[0029] The yaw rate dynamic equation characterizes the influence of the front axle lateral force and the rear axle lateral force on the vehicle's yaw motion, and the center-of-gravity sideslip angle dynamic equation characterizes the influence of the front axle lateral force and the rear axle lateral force on the vehicle's lateral motion. The yaw rate dynamic equation and the center-of-gravity sideslip angle dynamic equation form the basic dynamic model of the vehicle.
[0030] Obtain the lateral force of the front axle disturbance, superimpose the lateral force of the front axle disturbance onto the lateral force of the front axle to obtain the superimposed lateral force of the front axle, and obtain the disturbance yaw moment based on the lateral force of the front axle disturbance.
[0031] The disturbance yaw moment is incorporated into the yaw rate dynamic equation, and the influence of the front axle disturbance lateral force on the vehicle's lateral motion is incorporated into the centroid sideslip angle dynamic equation, resulting in the disturbance centroid sideslip angle dynamic equation. Through this process, the influence of the front axle disturbance lateral force is reflected in both the vehicle's yaw and lateral motion.
[0032] Introducing the rate of change of the disturbance centroid sideslip angle, the rate of change of the disturbance centroid sideslip angle is superimposed on the dynamic equation of the disturbance centroid sideslip angle to obtain the superimposed dynamic equation of the disturbance centroid sideslip angle.
[0033] The disturbance extended model is constructed from the superimposed disturbance centroid sideslip angle dynamic equation and the yaw rate dynamic equation after incorporating the disturbance yaw moment. The disturbance centroid sideslip angle change rate is used to characterize the effect of road roll on the lateral motion of the vehicle.
[0034] The yaw rate, the centroid sideslip angle, the disturbance yaw moment, and the rate of change of the disturbance centroid sideslip angle are combined to form an extended state vector.
[0035] Based on the perturbation extended model, state estimates are obtained using a perturbation observer. These state estimates include estimates of yaw rate, center of mass sideslip angle, perturbation yaw moment, and rate of change of perturbation center of mass sideslip angle.
[0036] Establish a preset conversion relationship between the rate of change of the disturbance centroid sideslip angle and the road side tilt angle, and extract the estimated value of the rate of change of the disturbance centroid sideslip angle from the state estimate value.
[0037] Based on the estimated rate of change of the disturbance centroid sideslip angle, the vehicle's longitudinal speed, and the preset conversion relationship, the estimated road tilt angle is calculated in reverse.
[0038] The beneficial effects of this embodiment are as follows: by introducing the lateral force of the front axle disturbance and the rate of change of the disturbance center of mass sideslip angle, the influence of vehicle model error and road roll on the lateral motion of the vehicle is characterized respectively. The yaw rate, center of mass sideslip angle, disturbance yaw moment and the rate of change of the disturbance center of mass sideslip angle are combined into an extended state vector for joint estimation, so that the road roll angle can be obtained from the disturbance state estimation results, thereby improving the accuracy of road roll angle estimation.
[0039] Preferably, in this embodiment, obtaining the front axle lateral force and rear axle lateral force of the vehicle, and establishing the vehicle's yaw rate dynamic equation and center-of-gravity sideslip angle dynamic equation based on the front axle lateral force and the rear axle lateral force, includes: During the vehicle development and calibration phase, the vehicle mass and yaw moment of inertia are calibrated, and the distances from the vehicle's center of gravity to the front axle and the rear axle are obtained based on the vehicle's wheelbase and center of gravity position. The calibrated vehicle mass... Vehicle yaw moment of inertia Distance from the vehicle's center of gravity to the front axle and the distance from the vehicle's center of gravity to the rear axle Write the vehicle parameter table that has been pre-configured in the vehicle controller.
[0040] When estimating the road roll angle, the vehicle mass is read from the vehicle parameter table. The vehicle's yaw moment of inertia The distance from the vehicle's center of gravity to the front axle and the distance from the vehicle's center of gravity to the rear axle .
[0041] Acquire longitudinal vehicle speed collected by vehicle sensors Front axle steering angle and yaw rate Extract the centroid sideslip angle from the extended state estimate of the current control cycle. And read the feedforward yaw moment acting on the vehicle from the control signal output by the vehicle control system. .
[0042] According to the yaw rate The centroid side slip angle and the longitudinal speed The system obtains the velocity directions of the front axle position and the rear axle position of the vehicle; it obtains the front axle slip angle based on the difference between the velocity direction of the front axle position and the front axle steering angle; and it obtains the rear axle slip angle based on the difference between the velocity direction of the rear axle position and the rear axle wheel direction.
[0043] The front axle slip angle and the rear axle slip angle are input into the Pacejka tire model, and the front axle lateral force is calculated using the Pacejka tire model. and rear axle lateral force .
[0044] According to the aforementioned front axle lateral force and the rear axle lateral force The yaw rate dynamic equation of the vehicle is established, and the yaw rate dynamic equation is expressed as: The dynamic equations for the vehicle's center of gravity sideslip angle are established, and these equations are expressed as follows: in, The yaw acceleration is... Indicates yaw rate. Indicates the centroid sideslip angle. This represents the yaw moment of inertia of the vehicle. Indicates vehicle mass. Indicates longitudinal vehicle speed, This indicates the distance from the vehicle's center of gravity to the front axle. This indicates the distance from the vehicle's center of gravity to the rear axle. Indicates the lateral force on the front axle. Indicates the lateral force on the rear axle. This represents the feedforward yaw moment acting on the vehicle.
[0045] like Figure 3 As shown, front axle lateral force Lateral forces acting on the front and rear axles of the vehicle Acting on the rear axle of the vehicle, the distance from the vehicle's center of gravity to the front axle is... The distance from the vehicle's center of gravity to the rear axle is .
[0046] Preferably, in this embodiment, the Pacejka tire model further includes: Because the relationship between tire lateral force and tire slip angle is nonlinear and cannot be directly characterized by the vehicle's fundamental dynamics equations, the Pacejka tire model is used to establish the relationship between front axle slip angle, rear axle slip angle, and front axle lateral force. Rear axle lateral force The nonlinear mapping relationship between them; the Pacejka tire model takes the front axle slip angle and rear axle slip angle as inputs, and the front axle lateral force and rear axle lateral force The output is used to replace the linear assumption between lateral force and sideslip angle in the fundamental dynamic equations, so that the front axle lateral force... and rear axle lateral force The calculation results can reflect the actual nonlinear mechanical characteristics of the tire.
[0047] The beneficial effects of adopting the above scheme are: by establishing a nonlinear mapping relationship between the slip angle and the lateral force through the Pacejka tire model, the calculation results of the front axle lateral force and the rear axle lateral force are closer to the real mechanical characteristics of the tire, providing a more accurate basic input for the subsequent dynamic equations and perturbation extension models based on the front axle lateral force and the rear axle lateral force.
[0048] Preferably, in this embodiment, the step of determining the disturbance yaw moment based on the front axle disturbance lateral force and incorporating the disturbance yaw moment into the yaw rate dynamic equation and the center of mass sideslip angle dynamic equation includes: Considering that the front axle sideslip angle is usually larger than the rear axle sideslip angle, the model error of the front axle lateral force is typically more significant when cornering. Furthermore, increasing the lateral force disturbance state variable would increase the computational burden of extended state estimation. Therefore, this embodiment only introduces the front axle disturbance lateral force into the front axle lateral force, while keeping the rear axle lateral force unchanged. The front axle disturbance lateral force serves as the sole lateral force disturbance and the model variable to be estimated, characterizing the deviation in the front axle lateral force model caused by tire nonlinearity and changes in road conditions.
[0049] The yaw rate dynamics equation describes the vehicle's rotational state about its center of mass, using torque as its dimension. However, the front axle disturbance lateral force uses force as its dimension. Because their dimensions are different, the front axle disturbance lateral force cannot be directly substituted into the yaw rate dynamics equation. Therefore, based on the aforementioned front axle disturbance lateral force... and the distance from the vehicle's center of gravity to the front axle The equivalent torque conversion is performed to obtain the disturbance yaw moment, which is expressed as: in, Indicates the lateral force of front axle disturbance. The yaw moment of disturbance generated around the vehicle's center of gravity.
[0050] The advantages of adopting the above scheme are: by setting the front axle disturbance lateral force as the only lateral force disturbance quantity, the part of the front axle lateral force that cannot be represented by the basic dynamic model can be used as a disturbance state for estimation; the disturbance yaw moment is obtained by converting the front axle disturbance lateral force into an equivalent torque based on the distance from the vehicle's center of gravity to the front axle, which provides a dimensionally consistent conversion basis for superimposing the disturbance quantity into the yaw rate dynamic equation.
[0051] Preferably, in this embodiment, the step of introducing the rate of change of the perturbation centroid sideslip angle into the centroid sideslip angle dynamic equation to construct the perturbation extended model includes: The disturbance yaw moment Superimposed on the aforementioned yaw rate dynamic equation, the superimposed yaw rate dynamic equation is obtained, which is expressed as follows: The dynamic equation for the center of mass sideslip angle has the dimension of lateral force and the disturbance yaw moment. Using torque as the dimension, the two have different dimensions, so the disturbance yaw moment cannot be directly superimposed into the dynamic equation of the center of gravity sideslip angle. It is based on the distance from the vehicle's center of gravity to the front axle. The disturbance yaw moment Equivalent lateral force converted to front axle position The equivalent lateral force is then superimposed onto the centroid sideslip angle dynamic equation to obtain the disturbed centroid sideslip angle dynamic equation, which is expressed as: The beneficial effect of adopting the above scheme is that the same disturbance yaw moment is superimposed on the yaw rate dynamic equation and the center of gravity sideslip angle dynamic equation according to two different dimensional conversion methods, so that the model deviation of the front axle lateral force can be reflected in the vehicle yaw motion and center of gravity sideslip motion at the same time, providing a dynamic basis for establishing the disturbance extended model.
[0052] Preferably, in this embodiment, the step of estimating the extended state vector using a perturbation observer based on the perturbation extension model to obtain a state estimate includes: When the road on which the vehicle is traveling has a road tilt angle, gravity generates a gravitational component along the road tilt direction. Based on this gravitational component and gravitational acceleration, and combined with the vehicle's longitudinal speed, the effect of road tilt on the vehicle's lateral motion is equivalently characterized to obtain the rate of change of the disturbance centroid sideslip angle. The rate of change of the disturbance centroid sideslip angle is expressed as: in, Represents gravitational acceleration. Indicates the road slope angle. This represents the gravitational component. This indicates the vehicle's longitudinal speed.
[0053] The rate of change of the side slip angle of the perturbation centroid Superimposed on the perturbation centroid sideslip angle dynamic equation, the superimposed perturbation centroid sideslip angle dynamic equation is obtained, which is expressed as: The superimposed yaw rate dynamic equation describes the vehicle's rotational state around its center of mass, and the superimposed perturbation center of mass sideslip dynamic equation describes the vehicle's sideslip motion. These are two coupled motion states in the vehicle's lateral motion and need to be predicted and estimated simultaneously through a unified observer architecture. Therefore, the superimposed yaw rate dynamic equation and the superimposed perturbation center of mass sideslip dynamic equation are arranged in rows to form the perturbation extended model.
[0054] State estimation is performed using a perturbation observer, specifically including: Based on the vehicle dynamics model, the state estimate of the previous moment, and the vehicle input signal, determine the extended state prediction value and the corresponding vehicle motion state prediction value at the current moment. Determine the observation deviation between the measured values of the vehicle's motion state and the corresponding predicted values of the vehicle's motion state; The extended state prediction value is corrected based on the observation bias and the observer gain to obtain the state estimate value at the current time.
[0055] Preferably, in this embodiment, the perturbation extension model includes: The disturbance yaw moment and the rate of change of the centroid side deflection angle of the disturbance The equivalent effects of tire lateral force model deviation and road roll on vehicle lateral motion are respectively characterized. Neither has a dynamic law that can be explicitly modeled; therefore, the disturbance yaw moment is... and the rate of change of the centroid side deflection angle of the disturbance The rate of change of all variables is set to zero, thus obtaining the perturbation extension model, which is expressed as: in, This represents the extended state vector composed of the yaw rate, the sideslip angle, the disturbance yaw moment, and the rate of change of the disturbance sideslip angle. This represents a nonlinear function composed of the superimposed yaw rate dynamic equation and the superimposed disturbance center of mass sideslip angle dynamic equation. Indicates the front axle steering angle. It represents the feedforward yaw moment acting on the vehicle; the two components with a value of zero represent the rate of change of the disturbance yaw moment and the rate of change of the disturbance centroid sideslip angle, respectively.
[0056] The beneficial effects of adopting the above scheme are as follows: the influence of road tilt on vehicle lateral motion is equivalently represented as the rate of change of the disturbance center of mass sideslip angle and superimposed on the dynamic equation of the center of mass sideslip angle, so that the influence of road tilt angle can be estimated by the observer in the form of dynamic state variables; for the two coupled motion states of yaw rate and center of mass sideslip angle, the corresponding two dynamic equations are arranged in rows to form a unified disturbance extension model; and for the case that there are no dynamic laws that can be explicitly modeled for both disturbance yaw moment and disturbance center of mass sideslip angle, the rate of change of both is set to zero, so that the disturbance extension model can be represented in the form of a complete nonlinear differential equation of extended state vector, providing a basis for subsequent state prediction and estimation.
[0057] Preferably, in this embodiment, the step of estimating the extended state vector using a perturbation observer based on the perturbation extension model to obtain a state estimate includes: Extract the centroid sideslip angle from the extended state estimate output from the previous control cycle. and yaw rate estimates Obtain the front axle steering angle collected by vehicle sensors. and longitudinal speed Furthermore, the velocity directions at the front and rear axle positions of the vehicle are obtained by superimposing the velocity direction of the vehicle's center of gravity with the additional rotational velocity component resulting from the product of the distance from each axle to the center of gravity and the yaw rate; based on the aforementioned center of gravity sideslip angle... The estimated yaw rate and the longitudinal speed The velocity directions of the vehicle's front axle and rear axle positions are obtained. Based on the velocity direction of the front axle position and the front axle steering angle... The difference between the two values is used to obtain the front axle slip angle; the difference between the velocity direction of the rear axle position and the wheel direction of the rear axle is used to obtain the rear axle slip angle.
[0058] The front axle slip angle and the rear axle slip angle are input into the Pacejka tire model, and the front axle lateral force is calculated using the Pacejka tire model. and rear axle lateral force .
[0059] Since the state estimate output in the previous control cycle only reflects the vehicle state in that cycle, and the vehicle state continues to change as the control cycle progresses, it is necessary to extrapolate the state estimate to the next control cycle based on vehicle dynamics to obtain state variables that match the actual vehicle state at the current moment; therefore, the front axle lateral force... The rear axle lateral force Extended state estimate of the current control cycle and feedforward yaw moment Input the disturbance extension model to perform state prediction, and obtain the predicted state quantity for the next control cycle. The predicted state quantity is expressed as: in, This represents a nonlinear function composed of the superimposed dynamic equations of yaw rate and the superimposed dynamic equations of the disturbance center of mass sideslip angle. This represents the predicted state quantity.
[0060] Based on the yaw rate component in the predicted state variables, the predicted yaw rate is obtained. The sideslip angle, disturbance yaw moment, and rate of change of disturbance sideslip angle in the predicted state variables, along with the front axle lateral force, rear axle lateral force, longitudinal vehicle speed, and feedforward yaw moment, are substituted into the dynamic equation of the disturbance sideslip angle to obtain the predicted rate of change of sideslip angle. The predicted rate of change of sideslip angle is added to the predicted yaw rate, and the sum is multiplied by the longitudinal vehicle speed to obtain the predicted lateral acceleration.
[0061] In this embodiment, the duration of the control cycle is used as the integration step size, and the nonlinear differential equation of the disturbance extension model is integrated in one step using the improved Euler method. First, a first state change rate is calculated based on the extended state estimate of the current control cycle and the vehicle input, and a temporary state quantity is obtained based on the first state change rate. Then, a second state change rate is calculated based on the temporary state quantity and the vehicle input, and a predicted state quantity for the next control cycle is obtained based on the first state change rate and the second state change rate.
[0062] The predicted state variables are obtained by one-step integration of the nonlinear differential equations of the perturbation extension model, without correction using measured signals collected by vehicle sensors. If there are modeling errors in the vehicle's basic dynamics model and the perturbation extension model, the predicted state variables will deviate from the vehicle's actual state. To measure the deviation between the predicted state variables and the vehicle's actual state, the measured yaw rate is used. Subtract the predicted yaw rate value The yaw rate residual was obtained; the measured lateral acceleration value was used. Subtract the predicted lateral acceleration value The lateral acceleration residual is obtained.
[0063] The predicted state variables, the yaw rate residual, and the lateral acceleration residual are input into a Romberg observer. The Romberg observer corrects the predicted state variables based on the yaw rate residual and the lateral acceleration residual, and outputs the state estimate for the next control cycle. The state estimate is expressed as: in, This represents the observer gain matrix.
[0064] The beneficial effects of adopting the above scheme are: the deviation between the measured signal collected by the vehicle sensor and the predicted state quantity is introduced into the observer correction process, so that the state estimate can continuously use the measured signal to correct the modeling error of the prediction model, which helps to avoid the accumulation of estimation error caused by relying solely on the prediction model.
[0065] Preferably, in this embodiment, before the Romberg observer outputs the state estimate for the next control cycle, the method further includes: Since the perturbation extension model is a nonlinear model, and there are no universal methods for determining observability and solving for the gain matrix in nonlinear systems, the nonlinear functions in the perturbation extension model... Taking the partial derivatives in each state variable direction yields the system matrix. and output matrix This completes the Jacobian linearization of the perturbation extension model, transforming the problem of observability determination and gain matrix solving for the nonlinear system into a linear system problem, resulting in a linear state-space model, which is expressed as: in, To expand the state vector The derivative with respect to time is the rate of change of the extended state vector; The output vector of the linear state-space model consists of yaw rate and lateral acceleration. The system matrix and the output matrix The elements are determined by the combination of vehicle mass, vehicle yaw moment of inertia, distance from vehicle center of gravity to front axle, distance from vehicle center of gravity to rear axle, longitudinal speed, and lateral stiffness of front and rear axles.
[0066] Linearization only transforms the problem into a linear system problem; it does not guarantee that the four state variables in the extended state vector can be reconstructed from the two output variables, yaw rate and lateral acceleration. If the four state variables cannot be completely reconstructed, the subsequently designed observer gain matrix will not be able to converge the state estimate to the true value. Therefore, the output matrix... The first and second rows are respectively denoted as and ,according to , , , The elements are stacked in order into four rows to construct an observability matrix. ,in, , They represent , With the system matrix The row vector obtained by multiplication; according to , , , The order refers to using these four row vectors sequentially as the observability matrix. The first, second, third, and fourth rows; when the rank of the observability matrix Q is equal to the dimension of the extended state vector, the linearized perturbation extension model is deemed to satisfy the observability condition. During the vehicle development and calibration phase, the observability matrix Q is calculated based on the vehicle mass, vehicle yaw moment of inertia, distance from the vehicle's center of mass to the front and rear axles, front and rear axle lateral stiffness, and longitudinal vehicle speed operating range, and the observability matrix Q is verified to be full rank.
[0067] After the observability condition is satisfied, directly apply the system matrix. and the output matrix Configure poles to solve the observer gain matrix This requires solving a system of quartic polynomial equations, which is difficult to obtain analytically. To circumvent this difficulty, the observability matrix is used... Multiplying the inverse matrix by the freely chosen parameters yields two transformation vectors. and and in accordance with , , , Arrange them in order to construct the coordinate transformation matrix. ,in, , To obtain the two column vectors, , The system matrix is represented respectively. and , The resulting column vectors from the multiplication; this order refers to using these four column vectors sequentially as the coordinate transformation matrix. The first column, the second column, the third column, and the fourth column.
[0068] Let the state vector of the observer in canonical coordinates be... Let the original extended state vector The state vector satisfies The transformed system matrix is denoted as... The output matrix is denoted as ,in: Through the coordinate transformation matrix For the system matrix and the output matrix By performing coordinate transformation, a canonical model of the observer with a fixed structural form is obtained. This decomposes the original problem of solving eight unknown elements coupled together into two uncoupled second-order subsystems: the yaw rate and the centroid sideslip angle coupled channel, and the perturbation yaw moment and the perturbation centroid sideslip angle change rate coupled channel. Each subsystem contains only two unknown elements and can be solved directly.
[0069] Based on the vehicle dynamics response speed, the observer's expected convergence speed, and the sensor noise level, observer expected poles with negative real parts are set for each of the two second-order subsystems. For zero eigenvalues formed by disturbance states with a rate of change of zero, observer expected poles with non-zero negative real parts are set. Expected characteristic polynomials are constructed based on the observer expected poles of each second-order subsystem, and the coefficients of these expected characteristic polynomials are matched with the coefficients of the characteristic polynomials of the corresponding subsystems in the observer canonical model to obtain the canonical gain matrix. .
[0070] The normalized gain matrix It is obtained in the normalized coordinate system after coordinate transformation and cannot be directly applied to the perturbation extension model in the original coordinate system. Therefore, the coordinate transformation matrix is used. With the canonical gain matrix Multiply by the normalized gain matrix Perform an inverse coordinate transformation to restore the original coordinate system, generating the observer gain matrix that can be directly used for state estimation of the perturbation extended model. The observer gain matrix Represented as: The advantages of adopting the above scheme are: the gain design problem of the nonlinear system is transformed into a linear system problem through Jacobi linearization; the observability matrix is used to verify whether the extended state vector can be reconstructed from the yaw rate and lateral acceleration; the solution of the observer gain matrix is transformed to the observer normalized coordinate system through coordinate transformation, and the normalized gain matrix is obtained through pole placement; the observer gain matrix used for state estimation of the perturbation extended model is obtained through inverse coordinate transformation, so that the observer error converges according to the set desired poles.
[0071] Preferably, in this embodiment, after obtaining the observer gain matrix, the desired lateral acceleration and desired yaw rate are determined using a single-track model. The dynamic driving coefficient is determined based on the deviation between the measured and desired values. The observer gain used to correct the rate of change of the disturbance center of mass sideslip angle is then adjusted using the dynamic driving coefficient. Specifically, this includes: The system acquires measured values of yaw rate, lateral acceleration, front axle steering angle, and longitudinal vehicle speed from vehicle sensors, and reads the feedforward yaw moment from the vehicle control system. It then reads the measured yaw rate values from the current control cycle and the previous control cycle, and divides the difference between the two yaw rate values by the control cycle duration to obtain the yaw acceleration.
[0072] During the vehicle development and calibration phase, the lateral forces generated by the front and rear axle tires at different slip angles are collected. Based on the relationship between lateral force and slip angle, the front and rear axle lateral stiffness are obtained. The front and rear axle lateral stiffness, vehicle yaw moment of inertia, distance from the vehicle's center of gravity to the front axle, and distance from the vehicle's center of gravity to the rear axle are then entered into a vehicle parameter table. When estimating road roll angle, these vehicle parameters are retrieved from the vehicle parameter table.
[0073] The method described in this embodiment for back-calculating the desired lateral acceleration and desired yaw rate using a single-track model refers to calculating the desired lateral acceleration and desired yaw rate corresponding to the current vehicle operating state based on vehicle structural parameters, front axle lateral stiffness, rear axle lateral stiffness, front axle steering angle, longitudinal vehicle speed, feedforward yaw moment, measured yaw rate, and measured lateral acceleration. The specific process includes the following steps: To calculate the desired lateral acceleration under the current vehicle operating conditions, the vehicle structural parameters and the front and rear axle lateral stiffness are first combined. The vehicle wheelbase is expressed as: The first intermediate parameter is obtained by subtracting the product of the distance from the vehicle's center of gravity to the front axle and the front axle lateral stiffness from the product of the distance from the vehicle's center of gravity to the rear axle and the rear axle lateral stiffness. The second intermediate parameter is obtained by multiplying the distance from the vehicle's center of gravity to the rear axle, the rear axle lateral stiffness, and the vehicle's wheelbase. The formula is: in, This indicates the distance from the vehicle's center of gravity to the front axle. This indicates the distance from the vehicle's center of gravity to the rear axle. Indicates the front axle lateral stiffness. This indicates the rear axle lateral stiffness.
[0074] The first and second intermediate parameters are used to convert vehicle structural parameters and front and rear axle lateral stiffness into coefficients required for lateral motion calculation. The first coefficient is obtained by dividing the sum of the front and rear axle lateral stiffness by the first intermediate parameter. Divide the second intermediate parameter by the first intermediate parameter to obtain the second coefficient. .
[0075] After processing the vehicle parameters, the vehicle operating parameters for the current control cycle are processed. The third intermediate parameter is obtained by subtracting the product of the vehicle's yaw moment of inertia and yaw angular acceleration from the feedforward yaw moment. .
[0076] The fourth intermediate parameter is obtained by subtracting the rear wheel steering angle from the tangent of the front axle steering angle, and then subtracting the ratio of the product of the vehicle wheelbase and the measured yaw rate to the longitudinal speed. .
[0077] When the vehicle does not have rear-wheel steering, the rear-wheel steering angle is set to zero. The first and second coefficients represent the influence of vehicle parameters on lateral motion, while the third and fourth intermediate parameters represent the vehicle's operating state in the current control cycle. Multiplying the first coefficient by the third intermediate parameter, and then multiplying the second coefficient by the fourth intermediate parameter, and summing the two products, yields the desired lateral acceleration. The formula for calculating the desired lateral acceleration is: in, Indicates the desired lateral acceleration. Indicates the first coefficient. Indicates the second coefficient. This represents the third intermediate parameter. This indicates the fourth intermediate parameter.
[0078] Next, the desired yaw rate is calculated by back-calculating the measured lateral acceleration value, and the third coefficient is obtained by dividing the first intermediate parameter by the second intermediate parameter. The fourth coefficient is obtained by dividing the sum of the front axle lateral stiffness and the rear axle lateral stiffness by the second intermediate parameter. .
[0079] The measured lateral acceleration value, the third coefficient, the fourth coefficient, the third intermediate parameter, the front axle steering angle, the rear wheel steering angle, and the longitudinal vehicle speed are combined according to the following formula to obtain the desired yaw rate, as follows: in, This represents the measured value of lateral acceleration. Indicates the desired lateral acceleration. This represents the desired yaw rate.
[0080] Therefore, the measured yaw rate is used in the inverse calculation of the desired lateral acceleration, and the measured lateral acceleration is used in the inverse calculation of the desired yaw rate, thus forming a lateral acceleration channel and a yaw rate channel.
[0081] The measured lateral acceleration is then subtracted from the desired lateral acceleration, and the absolute value of the subtraction is taken to obtain the lateral acceleration deviation. The deviation of the yaw rate is obtained by subtracting the expected yaw rate from the measured yaw rate and taking the absolute value of the subtraction. .
[0082] Since the lateral acceleration deviation and the yaw rate deviation have different dimensions and numerical ranges, the two deviations are first converted into dynamic driving coefficient components with the same value range, and then the two dynamic driving coefficient components are fused.
[0083] During the vehicle development and calibration phase, multiple speed ranges are divided according to longitudinal vehicle speed. Within each speed range, measured values of yaw rate and lateral acceleration are collected under stable driving conditions and dynamic driving conditions. The lateral acceleration deviation and yaw rate deviation are calculated according to the above back-calculation process.
[0084] Multiple lateral acceleration deviation calibration points are set according to the magnitude of the lateral acceleration deviation, and a coefficient value between zero and one is assigned to each lateral acceleration deviation calibration point to form a lateral acceleration channel characteristic curve. The lateral acceleration deviation obtained under stable driving conditions is assigned a coefficient value close to zero, and the lateral acceleration deviation obtained under dynamic driving conditions is assigned a coefficient value close to one.
[0085] Within each vehicle speed range, multiple yaw rate deviation calibration points are set according to the magnitude of the yaw rate deviation, and a coefficient value between zero and one is assigned to each calibration point, forming a yaw rate channel characteristic curve. The yaw rate deviation obtained under stable driving conditions is assigned a coefficient value close to zero, while the yaw rate deviation obtained under dynamic driving conditions is assigned a coefficient value close to one. The lateral acceleration channel characteristic curve and the yaw rate channel characteristic curve for each vehicle speed range are stored in the vehicle controller.
[0086] When estimating the road roll angle, two adjacent calibration points where the lateral acceleration deviation occurs are located in the lateral acceleration channel characteristic curve. The coefficient values configured for the two adjacent calibration points are then linearly interpolated to obtain the dynamic driving coefficient components of the lateral acceleration channel. ; Next, the current vehicle speed range is determined based on the longitudinal vehicle speed. Then, two adjacent calibration points where the yaw rate deviation occurs are located in the yaw rate channel characteristic curve of the current vehicle speed range. The coefficient values configured for these two adjacent calibration points are then linearly interpolated to obtain the dynamic driving coefficient component of the yaw rate channel. The calculation formula is: in, This represents the lateral acceleration channel characteristic curve that has been pre-calibrated and stored in the vehicle controller. This represents the yaw rate channel characteristic curve, calibrated according to longitudinal vehicle speed zones and stored in the vehicle controller. Both functions are lookup table relationships based on calibration points, eliminating the need for function solving during road roll angle estimation. The dynamic driving coefficient is obtained by minimizing the lateral acceleration channel dynamic driving coefficient component and the yaw rate channel dynamic driving coefficient component. The minimum value is selected so that the dynamic driving coefficient is simultaneously constrained by the lateral acceleration channel and the yaw rate channel. The dynamic driving coefficient will only increase when both channels indicate that the actual vehicle motion deviates from the desired motion state, thereby reducing false triggering caused by fluctuations in the signal from a single sensor.
[0087] After obtaining the dynamic driving coefficient, the yaw rate residual and lateral acceleration residual of the current control cycle are processed by the preset observer correction gain to obtain the preliminary correction amount of the disturbance centroid sideslip angle change rate.
[0088] The preliminary correction is superimposed onto the predicted value of the disturbance centroid sideslip angle change rate for the current control cycle to obtain a preliminary estimate of the disturbance centroid sideslip angle change rate for the current control cycle. This preliminary estimate is then compared with the state estimate of the disturbance centroid sideslip angle change rate output from the previous control cycle to obtain a cycle comparison result. When the preliminary estimate is greater than the state estimate of the disturbance centroid sideslip angle change rate from the previous control cycle, it indicates that the disturbance centroid sideslip angle change rate is increasing. At this point, the two gain elements in the preset observer gain matrix used to correct the disturbance centroid sideslip angle change rate are scaled using the dynamic driving coefficient: in, and This represents the scaled observer correction gain. The preset observer correction gain represents the rate of change of the yaw rate residual acting on the sideslip angle of the perturbation centroid. The preset observer correction gain represents the rate of change of the sideslip angle of the perturbation centroid due to the lateral acceleration residual. This represents the gain attenuation factor pre-written into the vehicle controller, and the value of the gain attenuation factor ranges from 0.5 to 1.
[0089] Furthermore, gain scaling only applies to the observer gain matrix. and The other gain elements in the observer gain matrix remain unchanged. The yaw rate residual and lateral acceleration residual of the current control cycle are reprocessed using the scaled observer correction gain to obtain the adjusted correction amount: The adjusted correction is superimposed on the predicted value of the disturbance centroid sideslip angle change rate for the current control period to obtain the state estimate of the disturbance centroid sideslip angle change rate for the current control period.
[0090] When the preliminary estimate is less than or equal to the state estimate of the rate of change of the disturbance centroid sideslip angle in the previous control cycle, the preset observer correction gain is not scaled. and The observer correction gain is used as the current control cycle, and the preliminary estimate is used as the state estimate of the rate of change of the disturbance centroid sideslip angle for the current control cycle.
[0091] Therefore, the current control cycle first forms a preliminary estimate by correcting the gain through a preset observer, and then selects whether to scale the observer to correct the gain by comparing the preliminary estimate with the state estimate of the previous control cycle. Finally, the state estimate of the disturbance centroid sideslip angle change rate of the current control cycle is output, forming a closed loop of processing for dynamic driving coefficient calculation, state change judgment, gain adjustment and state correction.
[0092] The beneficial effects of adopting the above scheme are as follows: the degree to which the actual vehicle response deviates from the expected response of the single-track model is identified by jointly using the lateral acceleration channel and the yaw rate channel, and the deviation of the two channels is converted into a dynamic driving coefficient with a uniform value range by using the pre-calibrated characteristic curve; when the rate of change of the disturbance center of gravity sideslip angle is increasing, the observer correction gain used to correct the rate of change of the disturbance center of gravity sideslip angle is reduced according to the dynamic driving coefficient, reducing the situation where the dynamic steering response is misjudged by the observer as the influence of road roll; when the rate of change of the disturbance center of gravity sideslip angle is not increasing, the preset observer correction gain is maintained, so that the estimated value of road roll angle can fall back normally.
[0093] Preferably, in this embodiment, before the Romberg observer outputs the state estimate for the next control cycle, the method further includes: When the vehicle's longitudinal speed decreases, the calculations related to the longitudinal speed in the disturbance extension model are prone to amplifying sensor noise and model errors, potentially causing fluctuations in the numerical integration results. To avoid calculating the critical stable speed of the disturbance extension model online, a vehicle speed threshold is directly set in the vehicle controller. In this embodiment, the vehicle speed threshold is set to 8 m / s.
[0094] Within each control cycle, the wheel speeds collected by the vehicle wheel speed sensors are acquired, and the vehicle's longitudinal speed is calculated based on the wheel speeds and wheel rolling radii; the front axle steering angle is read from the vehicle steering system. The vehicle's longitudinal speed is compared with a speed threshold to obtain a speed comparison result.
[0095] When the vehicle speed comparison result indicates that the longitudinal vehicle speed is greater than or equal to the vehicle speed threshold, the extended state vector is predicted using the perturbation extended model, and the predicted state quantity is corrected by the Romberg observer to obtain the extended state estimate for the current control cycle.
[0096] When the vehicle speed comparison result indicates that the vehicle's longitudinal speed is lower than the vehicle speed threshold, the Ackerman kinematics calculation process is initiated, and the extended state estimates output by the disturbance extension model and the Romberg observer are no longer used as the state estimates for the current control cycle.
[0097] During the Ackermann kinematics calculation process, the distances from the vehicle's center of gravity to the front axle and the distances from the vehicle's center of gravity to the rear axle are read from the vehicle parameter table. The two distances are summed to obtain the vehicle's wheelbase. In this embodiment, the rear wheels of the vehicle do not participate in steering, and the rear wheel plane is parallel to the vehicle's longitudinal direction.
[0098] At low vehicle speeds, the lateral slippage generated by the tires is relatively small, therefore the front and rear wheels of the vehicle are considered to be in a state of pure rolling. Pure rolling means that the direction of wheel movement is consistent with the direction of the wheel plane, and is used to determine the instantaneous center of rotation based on the vehicle's geometry.
[0099] First, determine if the front axle steering angle is zero. When the front axle steering angle is zero, the vehicle travels in a straight line along the longitudinal direction. The process of calculating the instantaneous center of rotation is no longer performed; instead, the kinematic yaw rate and the kinematic center of mass sideslip angle are directly set to zero.
[0100] When the front axle steering angle is not zero, a first geometric line is formed with the center of the rear axle as the first reference point and along a direction perpendicular to the rear wheel plane; a second geometric line is formed with the center of the front axle as the second reference point and along a direction perpendicular to the front wheel plane. The angle between the front wheel plane and the longitudinal direction of the vehicle is the front axle steering angle.
[0101] The intersection point of the first and second geometric lines is determined, and this intersection point is taken as the instantaneous rotation center of the vehicle within the current control cycle. The instantaneous rotation center represents the position of the vehicle about its current front axle steering angle.
[0102] Calculate the distance from the instantaneous rotation center to the vehicle's center of gravity to obtain the kinematic turning radius of the vehicle's center of gravity. Divide the vehicle's longitudinal speed by the kinematic turning radius to obtain the absolute value of the kinematic yaw rate; when the current axle steering angle is positive, set the kinematic yaw rate to a positive value; when the current axle steering angle is negative, set the kinematic yaw rate to a negative value.
[0103] After obtaining the instantaneous center of rotation, connect the vehicle's center of gravity and the instantaneous center of rotation to form a center of gravity radius line; draw a tangent perpendicular to the center of gravity radius line through the vehicle's center of gravity, and take the direction of the tangent line as the direction of motion of the vehicle's center of gravity. Calculate the angle between the direction of motion of the vehicle's center of gravity and the vehicle's longitudinal direction to obtain the kinematic center of gravity sideslip angle, and determine the direction of the kinematic center of gravity sideslip angle according to the sign of the front axle steering angle.
[0104] Therefore, the front axle steering angle and vehicle wheelbase are used to determine the instantaneous rotation center, which is used to determine the kinematic turning radius and the direction of the vehicle's center of mass motion. The vehicle's longitudinal speed divided by the kinematic turning radius yields the kinematic yaw rate, and the angle between the direction of the vehicle's center of mass motion and the vehicle's longitudinal direction forms the kinematic center of mass sideslip angle, thus obtaining the two kinematic state quantities at low vehicle speeds.
[0105] After obtaining the kinematic yaw rate and kinematic sideslip angle, the disturbance yaw moment and the rate of change of the disturbance sideslip angle are set to zero. A low-speed extended state vector is constructed according to the order of yaw rate, sideslip angle, disturbance yaw moment, and rate of change of the disturbance sideslip angle in the extended state vector. This low-speed extended state vector is written into the storage area used to store the extended state estimate for the current control cycle, overwriting the extended state estimates calculated by the disturbance extension model and the Romberg observer. The overwritten low-speed extended state vector is output as the low-speed state estimate for the current control cycle and saved as the initial extended state vector for the next control cycle.
[0106] Since the rate of change of the disturbance centroid sideslip angle in the low-speed extended state vector is zero, the road roll angle output value under low-speed conditions is set to zero to stop the output of invalid road roll angle estimation results caused by fluctuations in low-speed values. Setting the road roll angle output value to zero is part of the low-speed protection process and does not indicate that the actual roll angle of the road where the vehicle is located is zero.
[0107] If the vehicle's longitudinal speed in the next control cycle is still lower than the speed threshold, the front axle steering angle and vehicle longitudinal speed are reread, the instantaneous rotation center, kinematic yaw rate and kinematic center of mass sideslip angle are updated, and the disturbance yaw moment and the rate of change of disturbance center of mass sideslip angle are set to zero.
[0108] When the vehicle's longitudinal speed in the next control cycle is greater than or equal to the speed threshold, the low-speed extended state vector saved in the previous control cycle is used as the initial extended state vector of the disturbance extended model, and state prediction and Romberg observer correction are re-executed. The extended state estimate obtained after correction is used as the output of the current control cycle, and the use of Ackerman kinematics calculation results to overwrite the extended state vector is stopped, completing the switch from the low-speed kinematics estimation process to the disturbance observer estimation process.
[0109] The advantages of adopting the above scheme are as follows: when the vehicle's longitudinal speed is lower than the speed threshold, the kinematic yaw rate and kinematic centroid sideslip angle are obtained based on the vehicle's geometry, front axle steering angle, and longitudinal speed, avoiding the continued use of the numerically sensitive disturbance extension model at low speeds to estimate the vehicle's motion state; by setting the disturbance yaw moment and the rate of change of the disturbance centroid sideslip angle to zero, the accumulation of model errors and sensor noise at low speeds into the disturbance state variables is prevented; by saving the low-speed extended state vector and using it as the initial state when restoring the disturbance observer estimation, a processing closed loop is formed between low-speed state coverage, continuous updating, and normal speed recovery.
[0110] Example 2 This embodiment relates to a road roll angle estimation system based on a disturbance observer. For example... Figure 2 As shown, the system includes a basic dynamic model establishment module, a disturbance yaw moment establishment module, a disturbance extended model establishment module, an extended state estimation module, and a road side tilt angle calculation module.
[0111] The basic dynamics model building module is used to obtain the front axle lateral force and rear axle lateral force of the vehicle, and output the vehicle's yaw rate dynamic equation and center of mass sideslip angle dynamic equation. The calculation methods for the front axle lateral force and rear axle lateral force, as well as the establishment methods for the two dynamic equations, are the same as in Example 1.
[0112] The disturbance yaw moment establishment module is connected to the basic dynamic model establishment module and is used to receive the front axle lateral force, the rear axle lateral force, the yaw rate dynamic equation, and the center of mass sideslip angle dynamic equation. The disturbance yaw moment establishment module introduces the front axle disturbance lateral force into the front axle lateral force and outputs the disturbance yaw moment, the yaw rate dynamic equation superimposed with the disturbance yaw moment, and the disturbance center of mass sideslip angle dynamic equation.
[0113] The disturbance extension model establishment module is connected to the disturbance yaw moment establishment module. It receives the disturbance yaw moment and two dynamic equations, introduces the rate of change of the disturbance center-of-mass sideslip angle into the disturbance center-of-mass sideslip angle dynamic equation, and outputs the disturbance extension model. The disturbance extension model uses yaw angular velocity, center-of-mass sideslip angle, disturbance yaw moment, and rate of change of the disturbance center-of-mass sideslip angle as state variables.
[0114] The extended state estimation module is connected to the disturbance extension model establishment module. It receives the disturbance extension model and constructs an extended state vector from the yaw rate, center-of-gravity sideslip angle, disturbance yaw moment, and rate of change of the disturbance center-of-gravity sideslip angle. The extended state estimation module acquires measured values of the front axle steering angle, longitudinal vehicle speed, feedforward yaw moment, yaw rate, and lateral acceleration. It then uses the disturbance extension model and a Romberg observer to predict and correct the extended state vector, outputting a state estimate.
[0115] like Figure 4 As shown, the extended state estimation module includes a state prediction path and a state correction path. The state prediction path receives the extended state estimate, front axle steering angle, longitudinal vehicle speed, and feedforward yaw moment of the current control cycle, outputs the predicted state quantities for the next control cycle using the disturbance extended model, and obtains the predicted yaw rate and lateral acceleration values based on the predicted state quantities.
[0116] The state correction path receives measured values of yaw rate and lateral acceleration. The measured yaw rate is subtracted from the predicted yaw rate to obtain the yaw rate residual; the measured lateral acceleration is subtracted from the predicted lateral acceleration to obtain the lateral acceleration residual. The yaw rate residual and the lateral acceleration residual are processed by the observer gain matrix to form a state correction value. This state correction value is used to correct the predicted state value, obtaining the extended state estimate for the next control cycle.
[0117] Figure 4 In this system, the feedback connection, where the extended state estimate of the next control cycle is returned to the input of the state prediction path, indicates that the extended state estimate will participate in state prediction as the new extended state estimate for the current control cycle in the next control cycle. Thus, the state prediction path, the state correction path, and the feedback connection form a closed-loop state estimation structure that updates cycle by cycle.
[0118] The extended state estimation module is also used to adjust the observer correction gain according to the dynamic driving coefficient described in Embodiment 1, and when the longitudinal vehicle speed is lower than the preset vehicle speed threshold, to cover the extended state estimation value output by the disturbance extended model with the low vehicle speed extended state vector.
[0119] The road tilt angle calculation module is connected to the extended state estimation module and is used to receive the state estimation value, extract the disturbance centroid sideslip angle change rate estimate from the state estimation value, and output the road tilt angle estimate value according to the disturbance centroid sideslip angle change rate estimate value, longitudinal vehicle speed and preset conversion relationship.
[0120] The extended state estimation module is also connected to the basic dynamic model establishment module, and is used to feed back the estimated yaw rate and center of gravity sideslip angle of the current control cycle to the basic dynamic model establishment module to update the front axle lateral force and rear axle lateral force for the next control cycle. Thus, the five modules form a data processing closed loop of basic dynamic model establishment, disturbance extension, state prediction and correction, road roll angle calculation, and state feedback.
[0121] The system is deployed on the vehicle controller, and the five modules are implemented by a computer program stored in the vehicle controller and executed by the processor of the vehicle controller according to the above data processing relationship.
[0122] The beneficial effects of this embodiment are as follows: through data transmission and status feedback between modules, the front axle lateral force model error and road roll effect can be used as extended states to participate in joint estimation, and the road roll angle can be calculated based on the estimated value of the change rate of the disturbance centroid sideslip angle. This improves the accuracy and real-time performance of road roll angle estimation while avoiding the need to repeatedly set up independent estimation systems.
[0123] In this embodiment, the road roll angle estimation system is located in the vehicle controller. The basic dynamic model establishment module, the disturbance yaw moment establishment module, the disturbance extended model establishment module, the extended state estimation module, and the road roll angle calculation module can be implemented by a computer program stored in the vehicle controller and executed by the processor in the vehicle controller.
[0124] The vehicle controller acquires measured values of yaw rate, lateral acceleration, front axle steering angle, and longitudinal vehicle speed through vehicle sensors, and reads the feedforward yaw moment through the vehicle control system. The vehicle controller's memory stores vehicle mass, vehicle yaw moment of inertia, distance from vehicle center of gravity to front axle, distance from vehicle center of gravity to rear axle, front axle lateral stiffness, rear axle lateral stiffness, observer gain matrix, dynamic driving coefficient characteristic curve, and vehicle speed threshold.
[0125] The processor reads the vehicle parameters and vehicle operating parameters mentioned above, calls the basic dynamics model establishment module to calculate the front axle lateral force and rear axle lateral force, and establishes the yaw rate dynamic equation and the center of mass sideslip angle dynamic equation; calls the disturbance yaw moment establishment module to set the front axle disturbance lateral force and converts the front axle disturbance lateral force into a disturbance yaw moment; calls the disturbance extension model establishment module to introduce the disturbance center of mass sideslip angle change rate to form a disturbance extension model.
[0126] The processor further invokes the extended state estimation module, which uses the disturbance extended model to predict the extended state vector. Based on the yaw rate residual and lateral acceleration residual, the predicted state quantities are corrected to obtain state estimates for yaw rate, centroid sideslip angle, disturbance yaw moment, and the rate of change of disturbance centroid sideslip angle. The road roll angle calculation module reads the state estimate of the rate of change of disturbance centroid sideslip angle and the longitudinal vehicle speed, calculates the estimated road roll angle according to a preset conversion relationship, and outputs the estimated road roll angle to the vehicle condition monitoring system or vehicle stability control system.
[0127] This embodiment also provides a computer device, which includes a processor and a memory. The memory stores a computer program, vehicle parameters, an observer gain matrix, a dynamic driving coefficient characteristic curve, a vehicle speed threshold, and an extended state estimate from the previous control cycle.
[0128] A disturbance lateral force is introduced into the front axle lateral force to obtain the disturbance yaw moment; the disturbance centroid sideslip angle change rate is introduced to establish a disturbance extension model, obtain the front axle lateral force and rear axle lateral force of the vehicle, and establish the yaw rate dynamic equation and the centroid sideslip angle dynamic equation; when the processor executes the computer program, it obtains the front axle lateral force and rear axle lateral force of the vehicle, and establishes the yaw rate dynamic equation and the centroid sideslip angle dynamic equation; a disturbance lateral force is introduced into the front axle lateral force, and the disturbance yaw moment is determined based on the disturbance lateral force; the disturbance yaw moment and the disturbance centroid sideslip angle change rate are introduced into the vehicle dynamics equation to construct a disturbance extension model; based on the disturbance extension model, the extended state vector is estimated using a disturbance observer; based on the state estimate of the disturbance centroid sideslip angle change rate, the longitudinal vehicle speed, and a preset conversion relationship, the estimated road roll angle is calculated.
[0129] When the vehicle's longitudinal speed is lower than the speed threshold, the processor calculates the kinematic yaw rate and kinematic centroid sideslip angle according to the Ackermann kinematics relation, sets the disturbance yaw moment and the rate of change of the disturbance centroid sideslip angle to zero, and uses the formed low-speed extended state vector to cover the extended state estimate output by the disturbance extended model.
[0130] This embodiment also provides a computer-readable storage medium storing a computer program. When the computer program is executed by a processor, it causes the processor to perform the aforementioned road roll angle estimation method. The computer-readable storage medium can be a read-only memory, random access memory, flash memory, solid-state memory, or other non-volatile storage medium in a vehicle controller capable of storing a computer program.
[0131] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and represent a limitation on a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.
[0132] Those skilled in the art will recognize that this invention can be implemented as a system, method, or computer program product. Therefore, this invention can be specifically implemented in the following forms: it can be entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, this invention can also be implemented as a computer program product contained in one or more computer-readable media, which includes computer-readable program code.
[0133] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.
Claims
1. A method for estimating road roll angle based on a disturbance observer, characterized in that, The method includes: Obtain the lateral forces of the front axle and rear axle of the vehicle, and establish the yaw rate dynamic equation and the center of mass sideslip angle dynamic equation based on the lateral forces of the front axle and rear axle. Obtain the lateral force of the front axle disturbance, and determine the disturbance yaw moment based on the lateral force of the front axle disturbance; The disturbance yaw moment is introduced into the yaw angular velocity dynamic equation and the centroid sideslip angle dynamic equation, and the rate of change of the disturbance centroid sideslip angle is introduced into the centroid sideslip angle dynamic equation to construct a disturbance extension model; The yaw rate, the sideslip angle, the disturbance yaw moment, and the rate of change of the sideslip angle are used to form an extended state vector. Based on the disturbance extension model, the extended state vector is estimated using a disturbance observer to obtain the state estimate. A preset conversion relationship is established between the rate of change of the disturbance centroid sideslip angle and the road sideslip angle. The state estimate is converted according to the preset conversion relationship to obtain the estimated road sideslip angle.
2. The method according to claim 1, characterized in that, The step of determining the disturbance yaw moment based on the lateral force of the front axle disturbance and incorporating the disturbance yaw moment into the yaw rate dynamic equation and the center of mass sideslip angle dynamic equation includes: The lateral force of the front axle disturbance is taken as the only lateral force disturbance, and the distance from the vehicle's center of gravity to the front axle is obtained. The disturbance yaw moment is obtained by performing an equivalent torque conversion based on the front axle disturbance lateral force and the distance from the vehicle's center of gravity to the front axle, and then the disturbance yaw moment is superimposed on the yaw rate dynamic equation. The equivalent lateral force is obtained by converting the disturbance yaw moment and the distance from the vehicle's center of gravity to the front axle, and then the equivalent lateral force is superimposed on the dynamic equation of the center of gravity sideslip angle.
3. The method according to claim 1, characterized in that, The step of incorporating the rate of change of the disturbance centroid sideslip angle into the centroid sideslip angle dynamic equation to construct the disturbance extension model includes: Based on the gravitational component generated by road tilt, the vehicle's longitudinal speed, and gravitational acceleration, the influence of road tilt on the vehicle's lateral motion is equivalently characterized to obtain the rate of change of the disturbance centroid sideslip angle. The rate of change of the disturbance centroid sideslip angle is superimposed on the centroid sideslip angle dynamic equation; The disturbance extension model is composed of the yaw rate dynamic equation and the center of mass sideslip angle dynamic equation.
4. The method according to claim 1, characterized in that, The step of estimating the extended state vector using a perturbation observer based on the perturbation extension model to obtain state estimates includes: Based on the extended state estimate of the current control cycle, the front axle steering angle, and the longitudinal vehicle speed, the front axle lateral force and the rear axle lateral force are calculated using the Pacejka tire model. The front axle lateral force, the rear axle lateral force, the extended state estimate of the current control cycle, and the feedforward yaw moment are input into the disturbance extension model for state prediction to obtain the predicted state quantities for the next control cycle. Calculate the predicted yaw rate and predicted lateral acceleration based on the predicted state quantities; The yaw rate residual is obtained by subtracting the predicted yaw rate value from the measured yaw rate value; the lateral acceleration residual is obtained by subtracting the predicted lateral acceleration value from the measured lateral acceleration value. The predicted state variables, the yaw rate residual, and the lateral acceleration residual are input into the Romberg observer, which then outputs the state estimate for the next control cycle.
5. The method according to claim 4, characterized in that, Before the Romberg observer outputs the state estimate for the next control cycle, the following steps are also included: The perturbation extended model is subjected to Jacobian linearization to obtain the system matrix and output matrix; Observability analysis is performed based on the system matrix and the output matrix; when the observability analysis results show that the linearized perturbation extension model satisfies the observability condition, a coordinate transformation matrix is constructed. The system matrix and the output matrix are transformed using the coordinate transformation matrix to form a canonical observer model. By combining the observer canonical model and the observer's desired poles, a canonical gain matrix is obtained; The observer gain matrix is generated by performing an inverse coordinate transformation on the normalized gain matrix using the coordinate transformation matrix.
6. The method according to claim 5, characterized in that, After obtaining the observer gain matrix, the following is also included: The desired lateral acceleration and desired yaw rate are determined by using a single-track model. The dynamic driving coefficient is determined based on the deviation between the measured lateral acceleration and the desired lateral acceleration, and the deviation between the measured yaw rate and the desired yaw rate. The predicted value of the rate of change of the disturbance centroid sideslip angle in the current control cycle is corrected by using the preset observer correction gain, and a preliminary estimate is obtained; When the preliminary estimate is greater than the state estimate of the disturbance centroid sideslip angle change rate in the previous control cycle, the gain element used to correct the disturbance centroid sideslip angle change rate in the preset observer correction gain is scaled according to the dynamic driving coefficient, and the predicted value is corrected using the scaled gain element to obtain the state estimate of the disturbance centroid sideslip angle change rate in the current control cycle. When the preliminary estimate is less than or equal to the state estimate of the rate of change of the disturbance centroid sideslip angle in the previous control cycle, the preliminary estimate is used as the state estimate of the rate of change of the disturbance centroid sideslip angle in the current control cycle.
7. The method according to claim 1, characterized in that, The step of estimating the extended state vector using a perturbation observer based on the perturbation extension model includes: Compare the vehicle's longitudinal speed with a preset speed threshold; When the longitudinal speed of the vehicle is lower than the preset speed threshold, the kinematic yaw rate and the kinematic center of mass sideslip angle are calculated according to the front axle steering angle and the longitudinal speed of the vehicle, based on the Ackermann kinematics relationship. The low-speed extended state vector is composed of the kinematic yaw rate, the kinematic center of mass sideslip angle, the disturbance yaw moment assigned to zero, and the disturbance center of mass sideslip angle change rate assigned to zero. The low-speed extended state vector is replaced with the extended state vector estimated based on the disturbance extension model to obtain the low-speed state estimate.
8. A road roll angle estimation system based on a disturbance observer, characterized in that, The system includes: The basic dynamics model establishment module is used to obtain the front axle lateral force and the rear axle lateral force of the vehicle, and to establish the yaw rate dynamic equation and the center of mass sideslip angle dynamic equation based on the front axle lateral force and the rear axle lateral force. The disturbance yaw moment establishment module is used to introduce the front axle disturbance lateral force into the front axle lateral force, determine the disturbance yaw moment based on the front axle disturbance lateral force, and introduce the disturbance yaw moment into the yaw angular velocity dynamic equation and the center of mass sideslip angle dynamic equation. The perturbation extension model building module is used to introduce the rate of change of the perturbation centroid sideslip angle into the centroid sideslip angle dynamic equation to construct the perturbation extension model; An extended state estimation module is used to form an extended state vector from the yaw rate, the centroid sideslip angle, the disturbance yaw moment, and the rate of change of the disturbance centroid sideslip angle. Based on the disturbance extended model, the extended state vector is estimated using a disturbance observer to obtain a state estimate. The road tilt angle calculation module is used to establish a preset conversion relationship between the rate of change of the disturbance centroid sideslip angle and the road tilt angle, and to convert the state estimate value according to the preset conversion relationship to obtain the estimated value of the road tilt angle.
9. A computer device, characterized in that, It includes a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the road roll angle estimation method according to any one of claims 1 to 7.
10. A computer-readable storage medium, characterized in that, It stores a computer program that, when executed by a processor, implements the road roll angle estimation method according to any one of claims 1 to 7.
