Method and device for estimating a vehicle suspension state parameter

Abstract

The application provides a vehicle suspension state parameter estimation method and device, and relates to the technical field of intelligent control of vehicle suspension, and the method comprises the steps of: constructing an interactive state observer based on at least two physical characteristic parameters of a vehicle suspension; determining prior state prediction parameters corresponding to each sub-observer based on a previous model probability and a transition probability matrix corresponding to each sub-observer; updating the prior state prediction parameters of each sub-observer based on current observation data corresponding to each sub-observer to obtain posterior state estimation parameters of each sub-observer; adaptively determining a current forgetting factor corresponding to each sub-observer based on the current observation data, an observation matrix and the prior state prediction parameters of each sub-observer; and fusing the current model probability, the posterior state estimation parameters and the current forgetting factor corresponding to all sub-observers to obtain a current overall state parameter estimation value corresponding to the interactive state observer. The application can improve the estimation accuracy of the suspension state parameters.

Description

Technical Field

[0001] This invention relates to the field of intelligent control technology for vehicle suspension, and in particular to a method and apparatus for estimating vehicle suspension state parameters. Background Technology

[0002] The suspension is a key component of the vehicle's driving system. The main function of the suspension is to mitigate road impacts and dampen vehicle vibrations. Therefore, the vehicle's vibration response data is an important reference for intelligent suspension control. For vehicle suspension control systems, the calculation of suspension state quantities faces the following two problems: (1) The contradiction between accuracy and calculation cost: In actual control, there are many suspension state quantities. The cost of using high-precision sensors to calculate each suspension state quantity is high and difficult to promote on a large scale; (2) Some state quantities are difficult to calculate directly: Some state quantities are difficult to measure directly through sensors. For example, tire dynamic deformation is the vertical deformation of the tire during vehicle driving, which reflects the vehicle's tire contact characteristics and driving safety. It is one of the state quantities that must be referenced in suspension design and control. However, tire dynamic deformation cannot be directly and accurately measured by conventional sensors.

[0003] To address the aforementioned issues, existing technologies generally employ a suspension state observer based on the Kalman filter algorithm. This observer uses a recursive filtering method to combine the system's dynamic prediction data with actual measurement data to minimize the variance of the estimation error, thereby obtaining the optimal state estimation result.

[0004] However, the Kalman filter algorithm has high requirements for model adaptability. If there is a significant difference between the state-space model of the observed system and the state observer, it will inevitably weaken the observer's ability to approximate the true state of the system, thereby reducing prediction accuracy. In the suspension control process, the vehicle suspension system has time-varying and nonlinear characteristics. For example, axle load offset during vehicle movement will change the equivalent sprung mass of the suspension, and there are structural nonlinear phenomena in the vehicle suspension. Therefore, the traditional Kalman filter algorithm leads to a certain mismatch in the state-space equations between the observed system and the state observer, resulting in lower estimation accuracy of suspension state parameters, which in turn affects the suspension control effect. Summary of the Invention

[0005] This invention provides a method and apparatus for estimating vehicle suspension state parameters, which solves the problem in the prior art where the use of traditional Kalman filtering algorithm leads to a certain mismatch between the state space equations of the observed system and the state observer, resulting in low estimation accuracy of suspension state parameters and thus affecting the suspension control effect.

[0006] This invention provides a method for estimating vehicle suspension state parameters, comprising the following steps.

[0007] An interactive state observer is constructed based on all physical characteristic parameters of the vehicle suspension. The interactive state observer includes at least two sub-observers corresponding to each of the physical characteristic parameters.

[0008] Based on the previous model probability and transition probability matrix corresponding to each of the sub-observers, the prior state prediction parameters corresponding to each sub-observer are determined.

[0009] The prior state prediction parameters of each sub-observer are updated based on the current observation data corresponding to each sub-observer, so as to obtain the posterior state estimation parameters of each sub-observer.

[0010] Based on the prior state prediction parameters of each sub-observer, the transition probability matrix, and the previous model probability corresponding to the interactive sub-observer, the previous model probability of each sub-observer is updated to obtain the current model probability.

[0011] Based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer, the current forgetting factor corresponding to each sub-observer is adaptively determined. The forgetting factor is used to adjust the memory length of historical observation data for each sub-observer.

[0012] By fusing the current model probabilities corresponding to all sub-observers, the prior state prediction parameters, and the posterior state estimation parameters, the estimated current overall state parameters corresponding to the interactive state observer are obtained.

[0013] According to the vehicle suspension state parameter estimation method provided by the present invention, determining the current forgetting factor corresponding to each of the sub-observers based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer includes: Based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer, the innovation statistics corresponding to each sub-observer are determined; each innovation statistics is used to characterize the degree of adaptation of each sub-observer to the current observation; Based on the information statistics corresponding to each of the sub-observers, the current forgetting factor corresponding to each of the sub-observers is determined.

[0014] According to the vehicle suspension state parameter estimation method provided by the present invention, determining the current forgetting factor corresponding to each of the sub-observers based on the innovation statistics corresponding to each of the sub-observers includes: If the information statistic corresponding to each of the sub-observers is greater than the first preset threshold, the first preset forgetting factor is determined as the current forgetting factor corresponding to each of the observers; If the information statistics corresponding to each of the sub-observers are less than the second preset threshold, the second preset forgetting factor is determined as the current forgetting factor corresponding to each of the observers; the second preset threshold is less than the first preset threshold, and the second preset forgetting factor is greater than the first preset forgetting factor; If the information statistic corresponding to each of the sub-observers is less than or equal to the first preset threshold and greater than or equal to the second preset threshold, the current forgetting factor corresponding to each of the sub-observers is determined based on the previous forgetting factor corresponding to each of the sub-observers and the information statistic.

[0015] According to the vehicle suspension state parameter estimation method provided by the present invention, the current overall state parameter estimate includes the current overall estimate and the current overall covariance; the posterior state estimate includes the current posterior estimate and the current posterior error covariance. The method of fusing the current forgetting factor, the current model probability, and the posterior state estimation parameters corresponding to each of the sub-observers to obtain the current overall state parameter estimate corresponding to the interactive state observer includes: By fusing the current model probability and the current posterior estimate corresponding to each of the sub-observers, the current overall estimate corresponding to the interactive state observer is obtained; By integrating the overall estimate, the current model probability corresponding to each of the sub-observers, the current forgetting factor, the current posterior error covariance, and the current posterior estimate, the current overall covariance corresponding to the interactive state observer is obtained.

[0016] According to the vehicle suspension state parameter estimation method provided by the present invention, the step of updating the previous model probability of each sub-observer based on the prior state prediction parameters of each sub-observer, the transition probability matrix, and the previous model probability corresponding to the interactive sub-observer to obtain the current model probability includes: Based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer, the current maximum likelihood probability corresponding to each sub-observer is determined; the current maximum likelihood probability is used to characterize the observation matching degree of each sub-observer. The current model probability corresponding to each sub-observer is determined based on the current maximum likelihood probability corresponding to each sub-observer, the transition probability matrix, and the previous model probability corresponding to the interactive sub-observer.

[0017] According to the vehicle suspension state parameter estimation method provided by the present invention, the prior state prediction parameters include state prediction values ​​and prior error covariance. The step of determining the prior state prediction parameters for each sub-observer based on the previous model probability and transition probability matrix corresponding to each of the sub-observers includes: Based on the previous model probability and transition probability matrix of each of the sub-observers, the previous mixed initial state of the interactive state observer is determined; the previous mixed initial state includes the previous state optimal estimate and the previous estimate covariance matrix; Based on the previous state optimal estimate, determine the state prediction value corresponding to each of the sub-observers; Based on the previously estimated covariance matrix, the prior error covariance corresponding to each of the sub-observers is determined.

[0018] According to the vehicle suspension state parameter estimation method provided by the present invention, the posterior state estimation parameters include the current posterior estimated value and the current posterior error covariance; The step of updating the prior state prediction parameters of each sub-observer based on the current observation data corresponding to each sub-observer to obtain the posterior state estimation parameters of each sub-observer includes: Based on the current observation data and observation matrix of each sub-observer, determine the Kalman gain of each sub-observer; Based on the Kalman gain of each sub-observer and the current observation data, the state prediction value is estimated posteriorly to obtain the current posterior estimate value of each sub-observer. The prior error covariance is updated based on the Kalman gain of each sub-observer and the observation matrix to obtain the current posterior error covariance of each sub-observer.

[0019] According to the vehicle suspension state parameter estimation method provided by the present invention, the step of constructing an interactive state observer based on all physical characteristic parameters of the vehicle suspension includes: Based on all the physical characteristic parameters of the vehicle suspension, a two-degree-of-freedom model of the suspension is constructed. Determine the time-varying parameters of the target from all physical property parameters; Based on the range of change corresponding to the target time-varying parameter, determine the model transition probability and at least two suspension usage parameters corresponding to the target time-varying parameter; the model transition probability is used to determine the transition probability matrix. Based on the vertical dynamic equations corresponding to the two-degree-of-freedom model of the suspension and at least two suspension operating parameters, at least two sub-observers are constructed; The interactive state observer is constructed based on the at least two sub-observers.

[0020] According to the vehicle suspension state parameter estimation method provided by the present invention, the construction of at least two sub-observers based on the vertical dynamic equations corresponding to the two-degree-of-freedom suspension model and the suspension operating parameters includes: Based on the vertical dynamic equations corresponding to the two-degree-of-freedom suspension model, the state-space equations are determined. Based on the suspension usage parameters and the state space equations, at least two sub-observers are constructed.

[0021] The present invention also provides a vehicle suspension state parameter estimation device, comprising the following modules.

[0022] A construction module is used to construct an interactive state observer based on all physical characteristic parameters of the vehicle suspension. The interactive state observer includes at least two sub-observers corresponding to each of the physical characteristic parameters. The state prediction module is used to determine the prior state prediction parameters corresponding to each of the sub-observers based on the previous model probability and transition probability matrix corresponding to each of the sub-observers. The observation update module is used to update the prior state prediction parameters of each sub-observer based on the current observation data corresponding to each sub-observer, so as to obtain the posterior state estimation parameters of each sub-observer. The probability update module is used to update the previous model probability of each sub-observer based on the prior state prediction parameters of each sub-observer, the transition probability matrix, and the previous model probability corresponding to the interactive sub-observer, so as to obtain the current model probability. An adaptive determination module is used to adaptively determine the current forgetting factor corresponding to each of the sub-observers based on the current observation data, observation matrix and prior state prediction parameters of each of the sub-observers. The forgetting factor is used to adjust the memory length of each of the sub-observers for historical observation data. The fusion module is used to fuse the current model probabilities, the prior state prediction parameters, and the posterior state estimation parameters corresponding to all sub-observers to obtain the current overall state parameter estimate corresponding to the interactive state observer.

[0023] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the vehicle suspension state parameter estimation method as described above.

[0024] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the vehicle suspension state parameter estimation method as described above.

[0025] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the vehicle suspension state parameter estimation method as described above.

[0026] This invention provides a method and apparatus for estimating vehicle suspension state parameters. Based on all physical characteristic parameters of the vehicle suspension, an interactive state observer is constructed, comprising at least two sub-observers corresponding to each physical characteristic parameter. Prior state prediction parameters for each sub-observer are determined based on the previous model probability and transition probability matrix of each sub-observer at the previous time step. The prior state prediction parameters are updated based on the current observation data corresponding to each sub-observer to obtain the posterior state estimation parameters for each sub-observer. The previous model probability of each sub-observer is updated based on the prior state prediction parameters, transition probability matrix, and the previous model probability corresponding to the interactive sub-observer to obtain the current model probability. Based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer, a current forgetting factor is adaptively determined to adjust the memory length of each sub-observer for historical observation data. Finally, the current model probability, posterior state estimation parameters, and current forgetting factor corresponding to all sub-observers are fused to obtain the current overall state parameter estimate corresponding to the interactive state observer. In this invention, independent observations by sub-observers based on different physical characteristic parameters cover the possible operating modes of the vehicle suspension system, effectively encompassing its time-varying characteristics. Through adaptive fusion of the estimation results from all sub-observers, the current overall state parameter estimate of the interactive state observer is obtained. This solves the model mismatch problem caused by the time-varying and nonlinear characteristics of the vehicle suspension system, significantly improving the estimation accuracy of suspension state parameters. It provides reliable state feedback for the real-time calculation of optimal action forces by the suspension control system, thereby optimizing the vehicle's vertical dynamics control and laying the foundation for improving and optimizing vehicle safety, ride comfort, and handling characteristics. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0028] Figure 1 This is a flowchart illustrating the vehicle suspension state parameter estimation method provided in an embodiment of the present invention.

[0029] Figure 2 This is a schematic diagram of the suspension two-degree-of-freedom model provided in an embodiment of the present invention.

[0030] Figure 3This is a schematic diagram of the vehicle suspension state parameter estimation device provided in an embodiment of the present invention.

[0031] Figure 4 This is a schematic diagram of the structure of the electronic device provided in an embodiment of the present invention. Detailed Implementation

[0032] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0033] To address the problem in existing technologies where the traditional Kalman filter algorithm leads to a mismatch in state-space equations between the observed system and the state observer, resulting in low estimation accuracy of suspension state parameters and consequently affecting suspension control performance, this invention provides a method for estimating vehicle suspension state parameters. Figure 1 This is a flowchart illustrating the vehicle suspension state parameter estimation method provided in an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes steps 110 to 160.

[0034] Step 110: Based on all the physical characteristic parameters of the vehicle suspension, construct an interactive state observer, which includes at least two sub-observers corresponding to each of the physical characteristic parameters.

[0035] Specifically, all physical characteristic parameters of the vehicle suspension can be obtained through parameter calibration. These physical characteristic parameters refer to variable parameters that change with the vehicle's operating conditions, and there must be at least two of them. These physical characteristic parameters may include sprung element mass, unsprung element mass, suspension elastic element stiffness coefficient, sprung mass displacement, unsprung mass displacement, and road vibration amplitude, etc. After determining all the physical characteristic parameters of the vehicle suspension, a corresponding interactive state observer is constructed based on each physical characteristic parameter. The full name of this interactive state observer can be Interacting Multiple Model Adaptive Forgetting Factor Kalman Filter (IMMAFFKF). This interactive state observer includes multiple sub-observers corresponding to each physical characteristic parameter. Each sub-observer is a complete Kalman filter. The specific parameter values ​​of the target physical characteristic parameter are different for each group of sub-observers, but the parameter values ​​of other physical characteristic parameters are the same, and all sub-observers satisfy the Markov condition. For example, for the five sub-observers corresponding to the sprung element mass, the specific parameter values ​​of the sprung element mass for the five sub-observers are 200kg, 225kg, 250kg, 275kg, and 300kg, respectively. However, the specific parameter values ​​of the unsprung element mass, suspension elastic element stiffness coefficient, sprung mass displacement, unsprung mass displacement, and road amplitude for the five sub-observers are all the same. In this interactive state observer, multiple sub-observers corresponding to different physical characteristic parameters are used to iteratively approximate a suspension model that can describe the time-varying characteristics of the vehicle suspension. This allows the overall state parameter estimation to track changes in the suspension physical characteristic parameters, thus solving the mismatch problem caused by a single model.

[0036] Step 120: Based on the previous model probability and transition probability matrix corresponding to each of the sub-observers, determine the prior state prediction parameters corresponding to each of the sub-observers.

[0037] Specifically, the previous model probability characterizes the probability of each sub-observer occurring at the previous time step, i.e., the probability that each sub-observer is the true system model of the vehicle suspension. The model transition probability in the transition probability matrix characterizes the probability that sub-observer j transforms into sub-observer i at the previous time step, and the prior state prediction parameter characterizes the predicted value of each sub-observer based on the state estimate at the previous time step for the current time step. After constructing the interactive state observer, each sub-observer interacts with the model according to the previous model probability and the transition probability matrix, so that each sub-observer has fused the information of all sub-observers at the previous time step before making a new round of prediction. At the current time step, each sub-observer can make a state prediction based on the fused information from the previous time step, and the prior state prediction parameter corresponding to each sub-observer can be obtained. It should be noted that when the suspension control system undergoes model transformation, model interaction can transmit fused information to the transformed sub-observer (i.e., the real system model), enabling multiple sub-observers to track the transformation between models more smoothly. Compared to running multiple independent models in parallel, this embodiment of the invention can avoid estimation divergence caused by model transformation, thereby improving the accuracy of state estimation.

[0038] Step 130: Update the prior state prediction parameters of each sub-observer based on the current observation data corresponding to each sub-observer to obtain the posterior state estimation parameters of each sub-observer.

[0039] Specifically, after each sub-observer performs independent state prediction, each sub-observer performs independent observation to obtain the current observation data corresponding to each sub-observer. Based on the current observation data, the prior state prediction parameters of each sub-observer are updated to obtain more accurate posterior state estimation parameters for each sub-observer.

[0040] Step 140: Based on the prior state prediction parameters of each sub-observer, the transition probability matrix, and the previous model probability corresponding to the interactive sub-observer, update the previous model probability of each sub-observer to obtain the current model probability.

[0041] Specifically, after determining the prior state prediction parameters of each sub-observer, the observation matching degree between each sub-observer and the current target state is determined by combining the prior state prediction parameters, the transition probability matrix, and the previous model probability of the interacting sub-observers that interact with each sub-observer. Based on this observation matching degree, the previous model probability of each sub-observer is updated to obtain the current model probability corresponding to each sub-observer. This current model probability is used to characterize the update probability that each sub-observer is identified as a true system model after the observation at time k. The higher the observation matching degree, the greater the current model probability.

[0042] Step 150: Based on the current observation data, observation matrix and prior state prediction parameters of each sub-observer, adaptively determine the current forgetting factor corresponding to each sub-observer. The forgetting factor is used to adjust the memory length of historical observation data for each sub-observer.

[0043] Specifically, before fusing to obtain the current overall state estimate parameter value, the current forgetting factor is calculated for each sub-observer based on the current observation data, observation matrix and prior state prediction parameters of each sub-observer. The current forgetting factor is used to adjust the memory length of each sub-observer for historical observation data, that is, to adjust the weight of the current posterior error covariance, thereby adjusting the magnitude of the current overall covariance in the current overall state parameter estimate value.

[0044] In this embodiment of the invention, by determining a current forgetting factor for each sub-observer, each sub-observer can dynamically adjust its dependence on historical information. When the sub-observer has a high degree of matching with the current target state, increasing the current forgetting factor can make full use of historical information for smoothing filtering and suppressing noise. When the sub-observer has a low degree of matching with the current target state, decreasing the current forgetting factor can quickly track the transformation of the vehicle suspension system. This weakens the uncertainty of sub-observers with low matching degree during fusion, optimizes the overall uncertainty, and thus improves the estimation accuracy and reliability of the interactive state observer in time-varying and nonlinear systems.

[0045] Step 160: Integrate the current model probabilities corresponding to all sub-observers, the prior state prediction parameters, and the posterior state estimation parameters to obtain the current overall state parameter estimate corresponding to the interactive state observer.

[0046] Specifically, after determining the current model probability of each sub-observer, the current model probability, prior state prediction parameters, and posterior state estimation parameters of all sub-observers are fused to obtain the final current overall state parameter estimate of the interactive state observer at the current time. This current overall state parameter estimate can participate in the state prediction of each sub-observer at the next time, and this current overall state parameter estimate is more reliable than the estimate of any single sub-observer.

[0047] This invention provides a method for estimating vehicle suspension state parameters. Based on all physical characteristic parameters of the vehicle suspension, an interactive state observer is constructed, comprising at least two sub-observers corresponding to each physical characteristic parameter. Prior state prediction parameters for each sub-observer are determined based on the previous model probability and transition probability matrix of each sub-observer at the previous time step. The prior state prediction parameters are updated based on the current observation data corresponding to each sub-observer to obtain the posterior state estimation parameters for each sub-observer. The previous model probability of each sub-observer is updated based on the prior state prediction parameters, transition probability matrix, and the previous model probability corresponding to the interactive sub-observer to obtain the current model probability. Based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer, a current forgetting factor is adaptively determined to adjust the memory length of each sub-observer for historical observation data. Finally, the current model probability, posterior state estimation parameters, and current forgetting factor corresponding to all sub-observers are fused to obtain the current overall state parameter estimate corresponding to the interactive state observer. In this embodiment of the invention, independent observations by sub-observers based on different physical characteristic parameters cover the possible operating modes of the vehicle suspension system, effectively encompassing the time-varying characteristics of the vehicle suspension system. Through adaptive fusion of the estimation results from all sub-observers, the current overall state parameter estimate of the interactive state observer is obtained. This solves the model mismatch problem caused by the time-varying and nonlinear characteristics of the vehicle suspension system in a single model, significantly improving the estimation accuracy of suspension state parameters. It provides reliable state feedback for the real-time calculation of optimal action forces by the suspension control system, thereby optimizing the vehicle's vertical dynamics control and laying the foundation for improving and optimizing the vehicle's safety, ride comfort, and handling characteristics.

[0048] In one embodiment, constructing an interactive state observer based on all physical characteristic parameters of the vehicle suspension includes: Based on all the physical characteristic parameters of the vehicle suspension, a two-degree-of-freedom model of the suspension is constructed. Determine the time-varying parameters of the target from all physical property parameters; Based on the range of change corresponding to the target time-varying parameter, determine the model transition probability and at least two suspension usage parameters corresponding to the target time-varying parameter; the model transition probability is used to determine the transition probability matrix. Based on the vertical dynamic equations corresponding to the two-degree-of-freedom model of the suspension and at least two suspension operating parameters, at least two sub-observers are constructed; The interactive state observer is constructed based on the at least two sub-observers.

[0049] Specifically, after obtaining multiple physical characteristic parameters through parameter calibration, a two-degree-of-freedom suspension model is constructed based on these parameters. Taking a real quarter-vehicle model as an example, this quarter-vehicle model is the smallest formal unit of the vehicle, including the suspension and tires. By theoretically abstracting and simplifying this quarter-vehicle model, the corresponding two-degree-of-freedom suspension model can be obtained. Figure 2 This is a schematic diagram of the structure of a two-degree-of-freedom suspension model provided in an embodiment of the present invention. The two-degree-of-freedom suspension model is as follows: Figure 2 As shown, the two-degree-of-freedom suspension model includes sprung elements, unsprung elements, suspension elastic elements, actuator elements, and tire elastic elements. The mass of the sprung element corresponding to the sprung element is denoted as... The displacement of the spring-loaded mass is denoted as The mass of the unsprung element corresponding to the unsprung element is denoted as . The unsprung mass displacement is denoted as The stiffness coefficient of the suspension elastic element corresponding to the suspension elastic element is denoted as . The actuating force of the actuator element corresponding to the actuator element is denoted as . The stiffness coefficient of the tire elastic element corresponding to the tire elastic element is denoted as . Road vibration amplitude is recorded as .

[0050] After determining the physical characteristic parameters and constructing a two-degree-of-freedom suspension model, suspension state data corresponding to each physical characteristic parameter under common vehicle operating conditions are obtained. Then, based on the suspension state data, the variation range of each physical characteristic parameter under common vehicle operating conditions is determined, and based on this variation range, the classic suspension operating parameters for each physical characteristic parameter are determined, thereby determining the model transition probabilities corresponding to each suspension operating parameter. The model transition probabilities of all suspension operating parameters corresponding to each physical characteristic parameter are used to determine the transition probability matrix corresponding to each physical characteristic parameter.

[0051] Next, based on the principles of vehicle vertical dynamics, the vertical dynamic equations of the two-degree-of-freedom suspension model are determined. Based on these vertical dynamic equations and the suspension operating parameters, at least two sub-observers are constructed.

[0052] In one embodiment, the construction of at least two sub-observers based on the vertical dynamic equations corresponding to the two-degree-of-freedom suspension model and the suspension operating parameters includes: Based on the vertical dynamic equations corresponding to the two-degree-of-freedom suspension model, the state-space equations are determined. Based on the suspension usage parameters and the state space equations, at least two sub-observers are constructed.

[0053] Specifically, the vertical dynamic equations are generalized into state-space equations, and the operating parameters of each suspension are substituted into these state-space equations. The resulting state-space equations after substituting the operating parameters of the suspension are the respective sub-observers.

[0054] For example, with the mass of the spring element Unsprung component mass Suspension elastic element stiffness coefficient Tire elastic element stiffness coefficient And taking the sprung element mass as the target time-varying parameter (i.e., the target physical characteristic parameter), according to the typical working conditions commonly used by vehicles, the range of variation of the sprung element mass is [200, 300] kg. Based on this range, let the typical values ​​of the suspension operating parameters corresponding to the sprung element mass be 200 kg, 225 kg, 250 kg, 275 kg, and 300 kg, respectively. Referring to the normal distribution, the transition probability equation shown in equation (1) is constructed. Equation (1) is: .

[0055] in, The model transition probability is represented by the suspension usage parameters corresponding to the mass of the sprung element. By substituting these parameters into equation (1), the model transition probability corresponding to each suspension usage parameter can be obtained.

[0056] Based on the prior probability distribution (such as a normal distribution) of the sprung component mass, the model transition probabilities between typical suspension operating parameters are calculated, and a transition probability matrix is ​​constructed. This transition probability matrix characterizes the transition characteristics of the vehicle suspension system under different sprung mass conditions. It can be calibrated according to common vehicle operating conditions or determined based on historical data statistics. For example, five typical values, such as 200kg, 225kg, 250kg, 275kg, and 300kg, can be used to construct the transition probability matrix.

[0057] Then, based on the principle of vehicle vertical dynamics, the vertical dynamic equation of the suspension two-degree-of-freedom model shown in equation (2) is constructed, and equation (2) is: .

[0058] in, Indicates the displacement of the spring-loaded mass Find the second derivative. Indicates the displacement of unsprung mass. Find the second derivative.

[0059] The vertical dynamic equation shown in equation (2) is generalized to the state-space equation shown in equation (3), which is: .

[0060] in, ; ; ; ; ; ; ; ; Indicates the displacement of the spring-loaded mass Find the first derivative; Indicates the displacement of unsprung mass. Find the first derivative.

[0061] Different suspension parameters result in different state-space equations. Substituting the classic suspension parameters corresponding to the sprung mass into equation (3) yields five different state-space equations. These five state-space equations are defined as five sub-observers, and the difference between the five sub-observers lies in the specific parameter values ​​corresponding to the sprung mass. An interactive state observer is constructed based on the five sub-observers. Using the sprung mass acceleration as the observation, the observation matrix of each sub-observer is constructed, as shown in equation (4). Equation (4) is: .

[0062] It should be noted that, since vehicle suspension is a time-varying system, the state-space equations change during vehicle suspension operation, meaning the sub-observers change. The model transition probability is used to characterize the probability that the vehicle suspension system transitions from sub-observer j to sub-observer i. This transition probability matrix represents the physical properties of the vehicle suspension and can be designed based on the common operating conditions of the vehicle suspension.

[0063] It should be noted that the vehicle's common operating conditions can be unloaded, fully loaded, and half-loaded. The unloaded condition can determine the lower limit of the variation range of each physical characteristic parameter, the fully loaded condition can determine the upper limit of the variation range of each physical characteristic parameter, and the half-loaded condition can determine the interval value of the variation range of each physical characteristic parameter.

[0064] In one embodiment, the prior state prediction parameters include the state prediction value and the prior error covariance; The step of determining the prior state prediction parameters for each sub-observer based on the previous model probability and transition probability matrix corresponding to each of the sub-observers includes: Based on the previous model probability and transition probability matrix of each of the sub-observers, the previous mixed initial state of the interactive state observer is determined; the previous mixed initial state includes the previous state optimal estimate and the previous estimate covariance matrix; Based on the previous state optimal estimate, determine the state prediction value corresponding to each of the sub-observers; Based on the previously estimated covariance matrix, the prior error covariance corresponding to each of the sub-observers is determined.

[0065] Specifically, after determining the previous model probability and transition probability matrix corresponding to each sub-observer, the optimal state estimates of all sub-observers at time k-1 are weighted and fused using equation (5) to obtain the previous optimal state estimate in the previous mixed initial state generated by all sub-observers after model interaction. This previous optimal state estimate is used as the initial state for state prediction at time k (i.e., the current time). Equation (5) is as follows: .

[0066] in, This represents the optimal estimate of the previous state of sub-observer i after model interaction at time k-1. This represents the optimal state estimate of sub-observer j at time k-1. Let represent the mixed probability of the vehicle suspension system transitioning from sub-observer j to sub-observer i at time k-1, and , This represents the transition probability matrix of the vehicle suspension system from sub-observer j to sub-observer i. Let represent the probability of sub-observer j occurring at time k-1, and r represent the total number of sub-observers corresponding to the time-varying parameters of the target in the interactive state observer, where 0≤j≤r and 0≤i≤r.

[0067] Then, using equation (6), the estimated covariance matrices of all sub-observers at time k-1 are fused to obtain the previous estimated covariance matrix in the previous mixed initial state generated by all sub-observers after model interaction. Equation (6) is as follows: .

[0068] in, This represents the previous estimated covariance matrix corresponding to sub-observer i after model interaction at time k-1. Let represent the estimated covariance matrix of sub-observer j at time k-1. Let represent the forgetting factor of sub-observer j at time k-1, where k≥2.

[0069] Then, each sub-observer independently predicts the state based on the previous optimal state estimate, obtaining the state prediction value at time k as shown in equation (7), which is: .

[0070] in, This represents the state prediction value obtained by sub-observer i at time k. and Both represent the state equations of sub-observer i. This represents the optimal action force at time k-1.

[0071] Furthermore, prior error covariance is estimated based on the previously estimated covariance matrix, resulting in the prior error covariance at time k shown in equation (8), which is: .

[0072] in, This represents the prior error covariance estimated by sub-observer i at time k. State equations transpose, This represents the model noise of the sub-observer.

[0073] In one embodiment, the posterior state estimation parameters include the current posterior estimate and the current posterior error covariance; The step of updating the prior state prediction parameters of each sub-observer based on the current observation data corresponding to each sub-observer to obtain the posterior state estimation parameters of each sub-observer includes: Based on the current observation data and observation matrix of each sub-observer, determine the Kalman gain of each sub-observer; Based on the Kalman gain of each sub-observer and the current observation data, the state prediction value is estimated posteriorly to obtain the current posterior estimate value of each sub-observer. The prior error covariance is updated based on the Kalman gain of each sub-observer and the observation matrix to obtain the current posterior error covariance of each sub-observer.

[0074] Specifically, after each sub-observer performs independent observations, the current observation data is obtained. Then, using equation (9), the Kalman gain corresponding to each sub-observer is determined based on the current observation data and observation matrix of each sub-observer. Equation (9) is as follows: .

[0075] in, This represents the Kalman gain of sub-observer i at time k. The observation matrix corresponding to sub-observer i can be calculated according to equation (1). The observation matrix corresponding to sub-observer i The transpose of , where R represents observation noise.

[0076] After determining the Kalman gain, the state prediction value is estimated posteriorly based on the Kalman gain of each sub-observer, the current observation data, and the observation matrix using equation (10). The corrected current posterior estimate value of each sub-observer is obtained, and equation (10) is: .

[0077] in, Let represent the current posterior estimate of sub-observer i at time k. This represents the current observation data of the sub-observer at time k.

[0078] Using equation (11), the prior error covariance is updated based on the identity matrix, the Kalman gain of each sub-observer, and the observation matrix, to obtain the corrected current posterior error covariance. Equation (11) is as follows: .

[0079] in, This represents the current posterior error covariance of sub-observer i at time k. I Represents the identity matrix.

[0080] It should be noted that when each sub-observator performs independent observations, it can acquire the current observation data through the sensor array. For example, when the target time-varying parameter is the sprung mass, an accelerometer can be used to measure the sprung mass acceleration of the vehicle. Furthermore, since the data directly measured by the sensor array may be subject to noise interference and omissions, after obtaining the current observation data, preprocessing operations such as data correction and filtering / denoising can be performed to improve the accuracy and reliability of the current observation data.

[0081] In one embodiment, updating the previous model probability of each sub-observer based on the prior state prediction parameters of each sub-observer, the transition probability matrix, and the previous model probability corresponding to the interacting sub-observer to obtain the current model probability includes: Based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer, the current maximum likelihood probability corresponding to each sub-observer is determined; the current maximum likelihood probability is used to characterize the observation matching degree of each sub-observer. The current model probability corresponding to each sub-observer is determined based on the current maximum likelihood probability corresponding to each sub-observer, the transition probability matrix, and the previous model probability corresponding to the interactive sub-observer.

[0082] Specifically, in this embodiment of the invention, maximum likelihood estimation is used to update the model probability. That is, using equation (12), the current maximum likelihood probability corresponding to each sub-observer is calculated based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer. This current maximum likelihood probability is used to characterize the observation matching degree or similarity between each sub-observer and the current target state. Equation (12) is: .

[0083] in, Let represent the current maximum likelihood probability of sub-observer i at time k, and N represent the dimension of the current observation data, where N=1. , .

[0084] After determining the current maximum likelihood probability, the current model probability of each sub-observer is determined using equation (13). Equation (13) is: .

[0085] in, Let represent the current model probability of sub-observer i at time k. .

[0086] In one embodiment, determining the current forgetting factor corresponding to each of the sub-observers based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer includes: Based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer, the innovation statistics corresponding to each sub-observer are determined; each innovation statistics is used to characterize the degree of adaptation of each sub-observer to the current observation; Based on the information statistics corresponding to each of the sub-observers, the current forgetting factor corresponding to each of the sub-observers is determined.

[0087] Specifically, using equation (14), the innovation statistics of each sub-observer are calculated based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer. Equation (14) is as follows: .

[0088] in, The information statistics of sub-observer j at time k represent the degree of adaptation of sub-observer j to the current observation.

[0089] After determining the innovation statistics, the innovation statistics of each sub-observer are compared with a preset threshold to determine the current forgetting factor of each sub-observer.

[0090] In one embodiment, determining the current forgetting factor corresponding to each of the sub-observers based on the innovation statistics corresponding to each of the sub-observers includes: If the information statistic corresponding to each of the sub-observers is greater than the first preset threshold, the first preset forgetting factor is determined as the current forgetting factor corresponding to each of the observers; If the information statistics corresponding to each of the sub-observers are less than the second preset threshold, the second preset forgetting factor is determined as the current forgetting factor corresponding to each of the observers; the second preset threshold is less than the first preset threshold, and the second preset forgetting factor is greater than the first preset forgetting factor; If the information statistic corresponding to each of the sub-observers is less than or equal to the first preset threshold and greater than or equal to the second preset threshold, the current forgetting factor corresponding to each of the sub-observers is determined based on the previous forgetting factor corresponding to each of the sub-observers and the information statistic.

[0091] Specifically, after determining the innovation statistics of each sub-observer, equation (15) is used to compare the innovation statistics with a first preset threshold and a second preset threshold, respectively. The first preset threshold is the upper limit of the threshold, and the second preset threshold is the lower limit of the threshold, i.e., the first preset threshold is greater than the second preset threshold. If the innovation statistics are greater than the first preset threshold, the current forgetting factor is determined as the first preset forgetting factor, which is the lower limit of the forgetting factor. If the innovation statistics are less than the second preset threshold, the current forgetting factor is determined as the second preset forgetting factor, which is the upper limit of the forgetting factor. If the innovation statistics are greater than or equal to the second preset threshold and less than or equal to the first preset threshold, the current forgetting factor of each sub-observer is calculated based on the previous forgetting factor, the learning rate, and the target innovation energy. Equation (15) is: .

[0092] in, This represents the current forgetting factor of sub-observer j at time k. This represents the first preset threshold. This indicates the second preset threshold. This represents the first preset forgetting factor. This represents the second presupposed forgetting factor. Indicates the learning rate. It represents the target new energy.

[0093] In one embodiment, the current population state parameter estimate includes the current population estimate and the current population covariance; the posterior state estimate includes the current posterior estimate and the current posterior error covariance. The method of fusing the current forgetting factor, the current model probability, and the posterior state estimation parameters corresponding to each of the sub-observers to obtain the current overall state parameter estimate corresponding to the interactive state observer includes: By fusing the current model probability and the current posterior estimate corresponding to each of the sub-observers, the current overall estimate corresponding to the interactive state observer is obtained; By integrating the current overall estimate, the current model probability corresponding to each of the sub-observers, the current forgetting factor, the current posterior error covariance, and the current posterior estimate, the current overall covariance corresponding to the interactive state observer is obtained.

[0094] Specifically, after determining the current forgetting factor, the current model probabilities and current posterior estimates of all sub-observers are weighted and fused using equation (16) to obtain the current overall estimate of the interactive state observer, as shown in equation (16): .

[0095] in, This represents the current overall estimate of the interactive state observer at time k.

[0096] Using equation (17), based on the overall estimate, the current model probability of all sub-observers, the current forgetting factor, the current posterior error covariance, and the current posterior estimate, the current overall covariance corresponding to the interactive state observer is calculated. Equation (17) is as follows: .

[0097] in, This represents the current population covariance of the interactive state observer at time k.

[0098] Furthermore, to verify the state estimation accuracy of the vehicle suspension state parameter estimation method proposed in this embodiment, it was applied to the state observation process of the vehicle suspension system. The root mean square percentage error (RMSPE) was used as an evaluation index to quantitatively measure the deviation between the estimated value and the true value. Referring to the international standard ISO 8608, the road grades were set as the common urban road grades A, B, C, and D. Simultaneously, two sets of control experiments were set up: a classical Kalman filter (KF, assuming a fixed sprung mass of 200 kg) and a classical interactive multiple model Kalman filter (IMMKF, without an adaptive forgetting factor mechanism) to highlight the technical advantages of this invention in dealing with the time-varying and nonlinear characteristics of vehicle suspension systems. The error comparison results of each experiment are shown in Table 1.

[0099] Table 1

[0100] The vehicle suspension state parameter estimation device provided by the present invention is described below. The vehicle suspension state parameter estimation device described below and the vehicle suspension state parameter estimation method described above can be referred to in correspondence.

[0101] This invention also provides a vehicle suspension state parameter estimation device. Figure 3 This is a schematic diagram of the vehicle suspension state parameter estimation device provided in an embodiment of the present invention, as shown below. Figure 3 As shown, the vehicle suspension state parameter estimation device 300 includes: a construction module 310, a state prediction module 320, an observation update module 330, a probability update module 340, an adaptive determination module 360, and a fusion module 360.

[0102] The construction module 310 is used to construct an interactive state observer based on all physical characteristic parameters of the vehicle suspension. The interactive state observer includes at least two sub-observers corresponding to each of the physical characteristic parameters.

[0103] The state prediction module 320 is used to determine the prior state prediction parameters corresponding to each of the sub-observers based on the previous model probability and transition probability matrix corresponding to each of the sub-observers.

[0104] The observation update module 330 is used to update the prior state prediction parameters of each sub-observer based on the current observation data corresponding to each sub-observer, so as to obtain the posterior state estimation parameters of each sub-observer.

[0105] The probability update module 340 is used to update the previous model probability of each sub-observer based on the prior state prediction parameters of each sub-observer, the transition probability matrix, and the previous model probability corresponding to the interactive sub-observer, so as to obtain the current model probability.

[0106] The adaptive determination module 350 is used to adaptively determine the current forgetting factor corresponding to each of the sub-observers based on the current observation data, observation matrix and prior state prediction parameters of each of the sub-observers. The forgetting factor is used to adjust the memory length of each of the sub-observers for historical observation data.

[0107] The fusion module 360 ​​is used to fuse the current model probabilities corresponding to all sub-observers, the prior state prediction parameters, and the posterior state estimation parameters to obtain the current overall state parameter estimate corresponding to the interactive state observer.

[0108] This invention provides a vehicle suspension state parameter estimation device. Based on all physical characteristic parameters of the vehicle suspension, an interactive state observer is constructed, comprising at least two sub-observers corresponding to each physical characteristic parameter. Prior state prediction parameters for each sub-observer are determined based on the previous model probability and transition probability matrix of each sub-observer at the previous moment. The prior state prediction parameters are updated based on the current observation data corresponding to each sub-observer to obtain the posterior state estimation parameters for each sub-observer. The previous model probability of each sub-observer is updated based on the prior state prediction parameters, transition probability matrix, and the previous model probability corresponding to the interactive sub-observer to obtain the current model probability. Based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer, a current forgetting factor is adaptively determined to adjust the memory length of each sub-observer for historical observation data. Finally, the current model probability, posterior state estimation parameters, and current forgetting factor corresponding to all sub-observers are fused to obtain the current overall state parameter estimate value corresponding to the interactive state observer. In this embodiment of the invention, independent observations by sub-observers based on different physical characteristic parameters cover the possible operating modes of the vehicle suspension system, effectively encompassing the time-varying characteristics of the vehicle suspension system. Through adaptive fusion of the estimation results from all sub-observers, the current overall state parameter estimate of the interactive state observer is obtained. This solves the model mismatch problem caused by the time-varying and nonlinear characteristics of the vehicle suspension system in a single model, significantly improving the estimation accuracy of suspension state parameters. It provides reliable state feedback for the real-time calculation of optimal action forces by the suspension control system, thereby optimizing the vehicle's vertical dynamics control and laying the foundation for improving and optimizing the vehicle's safety, ride comfort, and handling characteristics.

[0109] Optionally, module 310 is specifically used for: Based on all the physical characteristic parameters of the vehicle suspension, a two-degree-of-freedom model of the suspension is constructed. Determine the time-varying parameters of the target from all physical property parameters; Based on the range of change corresponding to the target time-varying parameter, determine the model transition probability and at least two suspension usage parameters corresponding to the target time-varying parameter; the model transition probability is used to determine the transition probability matrix. Based on the vertical dynamic equations corresponding to the two-degree-of-freedom model of the suspension and at least two suspension operating parameters, at least two sub-observers are constructed; The interactive state observer is constructed based on the at least two sub-observers.

[0110] Optionally, module 310 is specifically used for: Based on the vertical dynamic equations corresponding to the two-degree-of-freedom suspension model and the suspension operating parameters, at least two sub-observers are constructed, including: Based on the vertical dynamic equations corresponding to the two-degree-of-freedom suspension model, the state-space equations are determined. Based on the suspension usage parameters and the state space equations, at least two sub-observers are constructed.

[0111] Optionally, the prior state prediction parameters include the state prediction value and the prior error covariance.

[0112] Optionally, the state prediction module 320 is specifically used for: Based on the previous model probability and transition probability matrix of each of the sub-observers, the previous mixed initial state of the interactive state observer is determined; the previous mixed initial state includes the previous state optimal estimate and the previous estimate covariance matrix; Based on the previous state optimal estimate, determine the state prediction value corresponding to each of the sub-observers; Based on the previously estimated covariance matrix, the prior error covariance corresponding to each of the sub-observers is determined.

[0113] Optionally, the posterior state estimation parameters include the current posterior estimate and the current posterior error covariance.

[0114] Optionally, the observation update module 330 is specifically used for: Based on the current observation data and observation matrix of each sub-observer, determine the Kalman gain of each sub-observer; Based on the Kalman gain of each sub-observer and the current observation data, the state prediction value is estimated posteriorly to obtain the current posterior estimate value of each sub-observer. The prior error covariance is updated based on the Kalman gain of each sub-observer and the observation matrix to obtain the current posterior error covariance of each sub-observer.

[0115] Optionally, the probability update module 340 is specifically used for: Based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer, the current maximum likelihood probability corresponding to each sub-observer is determined; the current maximum likelihood probability is used to characterize the observation matching degree of each sub-observer. The current model probability corresponding to each sub-observer is determined based on the current maximum likelihood probability corresponding to each sub-observer, the transition probability matrix, and the previous model probability corresponding to the interactive sub-observer.

[0116] Optionally, the adaptive determination module 350 is specifically used for: Based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer, the innovation statistics corresponding to each sub-observer are determined; each innovation statistics is used to characterize the degree of adaptation of each sub-observer to the current observation; Based on the information statistics corresponding to each of the sub-observers, the current forgetting factor corresponding to each of the sub-observers is determined.

[0117] Optionally, the adaptive determination module 350 is specifically used for: If the information statistic corresponding to each of the sub-observers is greater than the first preset threshold, the first preset forgetting factor is determined as the current forgetting factor corresponding to each of the observers; If the information statistics corresponding to each of the sub-observers are less than the second preset threshold, the second preset forgetting factor is determined as the current forgetting factor corresponding to each of the observers; the second preset threshold is less than the first preset threshold, and the second preset forgetting factor is greater than the first preset forgetting factor; If the information statistic corresponding to each of the sub-observers is less than or equal to the first preset threshold and greater than or equal to the second preset threshold, the current forgetting factor corresponding to each of the sub-observers is determined based on the previous forgetting factor corresponding to each of the sub-observers and the information statistic.

[0118] Optionally, the current population state parameter estimate includes the current population estimate and the current population covariance; the posterior state estimate includes the current posterior estimate and the current posterior error covariance.

[0119] Optionally, the fusion module 360 ​​is specifically used for: By fusing the current model probability and the current posterior estimate corresponding to each of the sub-observers, the current overall estimate corresponding to the interactive state observer is obtained; By integrating the overall estimate, the current model probability corresponding to each of the sub-observers, the current forgetting factor, the current posterior error covariance, and the current posterior estimate, the current overall covariance corresponding to the interactive state observer is obtained.

[0120] Figure 4 This is a schematic diagram of the structure of the electronic device provided in the embodiment of the present invention, such as... Figure 4As shown, the electronic device may include: a processor 410, a communication interface 420, a memory 430, and a communication bus 440, wherein the processor 410, the communication interface 420, and the memory 430 communicate with each other through the communication bus 440. The processor 410 can call logical instructions in the memory 430 to execute a vehicle suspension state parameter estimation method, which includes: constructing an interactive state observer based on all physical characteristic parameters of the vehicle suspension, the interactive state observer including at least two sub-observers corresponding to each physical characteristic parameter; determining the prior state prediction parameters corresponding to each sub-observer based on the previous model probability and transition probability matrix corresponding to each of the sub-observers; updating the prior state prediction parameters of each sub-observer based on the current observation data corresponding to each sub-observer to obtain the posterior state estimation parameters of each sub-observer; and updating the prior state prediction parameters of each sub-observer based on the current observation data corresponding to each sub-observer. The prior state prediction parameters of the observer, the transition probability matrix, and the previous model probability corresponding to the interactive sub-observer are used to update the previous model probability of each sub-observer to obtain the current model probability. Based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer, the current forgetting factor corresponding to each sub-observer is adaptively determined. The forgetting factor is used to adjust the memory length of each sub-observer for historical observation data. The current model probability, prior state prediction parameters, and posterior state estimation parameters corresponding to all sub-observers are fused to obtain the current overall state parameter estimate corresponding to the interactive state observer.

[0121] Furthermore, the logical instructions in the aforementioned memory 430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0122] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer can execute the vehicle suspension state parameter estimation method provided by the above methods. The method includes: constructing an interactive state observer based on all physical characteristic parameters of the vehicle suspension, wherein the interactive state observer includes at least two sub-observers corresponding to each of the physical characteristic parameters; determining the prior state prediction parameters corresponding to each of the sub-observers based on the previous model probability and transition probability matrix corresponding to each of the sub-observers; and performing prior state prediction on each of the sub-observers based on the current observation data corresponding to each of the sub-observers. The parameters are updated to obtain the posterior state estimation parameters of each sub-observer; based on the prior state prediction parameters of each sub-observer, the transition probability matrix, and the previous model probability corresponding to the interactive sub-observer, the previous model probability of each sub-observer is updated to obtain the current model probability; based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer, the current forgetting factor corresponding to each sub-observer is adaptively determined, and the forgetting factor is used to adjust the memory length of each sub-observer for historical observation data; the current model probability, prior state prediction parameters, and posterior state estimation parameters corresponding to all sub-observers are fused to obtain the current overall state parameter estimate value corresponding to the interactive state observer.

[0123] In another aspect, the present invention also provides a non-transitory computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements a vehicle suspension state parameter estimation method provided by the methods described above. This method includes: constructing an interactive state observer based on all physical characteristic parameters of the vehicle suspension, the interactive state observer including at least two sub-observers corresponding to each of the physical characteristic parameters; determining prior state prediction parameters corresponding to each of the sub-observers based on the previous model probability and transition probability matrix corresponding to each of the sub-observers; and updating the prior state prediction parameters of each of the sub-observers based on the current observation data corresponding to each of the sub-observers, thereby obtaining the prior state prediction parameters of each of the sub-observers. The posterior state estimation parameters of the observer are obtained; based on the prior state prediction parameters of each sub-observer, the transition probability matrix, and the previous model probability corresponding to the interactive sub-observer, the previous model probability of each sub-observer is updated to obtain the current model probability; based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer, the current forgetting factor corresponding to each sub-observer is adaptively determined, and the forgetting factor is used to adjust the memory length of each sub-observer for historical observation data; the current model probability, the prior state prediction parameters, and the posterior state estimation parameters corresponding to all sub-observers are fused to obtain the current overall state parameter estimate value corresponding to the interactive state observer.

[0124] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0125] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0126] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for estimating vehicle suspension state parameters, characterized in that, include: An interactive state observer is constructed based on all physical characteristic parameters of the vehicle suspension. The interactive state observer includes at least two sub-observers corresponding to each of the physical characteristic parameters. Based on the previous model probability and transition probability matrix corresponding to each of the sub-observers, determine the prior state prediction parameters corresponding to each of the sub-observers. Based on the current observation data corresponding to each sub-observer, the prior state prediction parameters of each sub-observer are updated to obtain the posterior state estimation parameters of each sub-observer. Based on the prior state prediction parameters of each sub-observer, the transition probability matrix, and the previous model probability corresponding to the interactive sub-observer, the previous model probability of each sub-observer is updated to obtain the current model probability; Based on the current observation data, observation matrix and prior state prediction parameters of each sub-observer, the current forgetting factor corresponding to each sub-observer is adaptively determined. The forgetting factor is used to adjust the memory length of each sub-observer for historical observation data. By fusing the current model probabilities corresponding to all sub-observers, the posterior state estimation parameters, and the current forgetting factor, the current overall state parameter estimate corresponding to the interactive state observer is obtained.

2. The vehicle suspension state parameter estimation method according to claim 1, characterized in that, The adaptive determination of the current forgetting factor for each sub-observer based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer includes: Based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer, the innovation statistics corresponding to each sub-observer are determined; each innovation statistics is used to characterize the degree of adaptation of each sub-observer to the current observation; Based on the information statistics corresponding to each of the sub-observers, the current forgetting factor corresponding to each of the sub-observers is determined.

3. The vehicle suspension state parameter estimation method according to claim 2, characterized in that, The step of determining the current forgetting factor for each sub-observer based on the innovation statistics corresponding to each sub-observer includes: If the information statistic corresponding to each of the sub-observers is greater than the first preset threshold, the first preset forgetting factor is determined as the current forgetting factor corresponding to each of the observers; If the information statistics corresponding to each of the sub-observers are less than the second preset threshold, the second preset forgetting factor is determined as the current forgetting factor corresponding to each of the observers; the second preset threshold is less than the first preset threshold, and the second preset forgetting factor is greater than the first preset forgetting factor. If the information statistic corresponding to each of the sub-observers is less than or equal to the first preset threshold and greater than or equal to the second preset threshold, the current forgetting factor corresponding to each of the sub-observers is determined based on the previous forgetting factor corresponding to each of the sub-observers and the information statistic.

4. The vehicle suspension state parameter estimation method according to claim 1, characterized in that, The current population state parameter estimate includes the current population estimate and the current population covariance; the posterior state estimate includes the current posterior estimate and the current posterior error covariance. The method of fusing the current model probabilities corresponding to all sub-observers, the posterior state estimation parameters, and the current forgetting factor to obtain the current overall state parameter estimate corresponding to the interactive state observer includes: By fusing the current model probability and the current posterior estimate corresponding to each of the sub-observers, the current overall estimate corresponding to the interactive state observer is obtained; By integrating the overall estimate, the current model probability corresponding to each of the sub-observers, the current forgetting factor, the current posterior error covariance, and the current posterior estimate, the current overall covariance corresponding to the interactive state observer is obtained.

5. The method for estimating vehicle suspension state parameters according to any one of claims 1-4, characterized in that, The method of updating the previous model probability of each sub-observer based on the prior state prediction parameters of each sub-observer, the transition probability matrix, and the previous model probability corresponding to the interactive sub-observer, to obtain the current model probability, includes: Based on the current observation data, observation matrix, and prior state prediction parameters of each sub-observer, the current maximum likelihood probability corresponding to each sub-observer is determined; the current maximum likelihood probability is used to characterize the observation matching degree of each sub-observer. The current model probability corresponding to each sub-observer is determined based on the current maximum likelihood probability corresponding to each sub-observer, the transition probability matrix, and the previous model probability corresponding to the interactive sub-observer.

6. The method for estimating vehicle suspension state parameters according to any one of claims 1-4, characterized in that, The prior state prediction parameters include the state prediction value and the prior error covariance; The step of determining the prior state prediction parameters for each sub-observer based on the previous model probability and transition probability matrix corresponding to each of the sub-observers includes: Based on the previous model probability and transition probability matrix of each of the sub-observers, the previous mixed initial state of the interactive state observer is determined; the previous mixed initial state includes the previous state optimal estimate and the previous estimate covariance matrix; Based on the previous state optimal estimate, determine the state prediction value corresponding to each of the sub-observers; Based on the previously estimated covariance matrix, the prior error covariance corresponding to each of the sub-observers is determined.

7. The vehicle suspension state parameter estimation method according to claim 6, characterized in that, The posterior state estimation parameters include the current posterior estimate and the current posterior error covariance; The step of updating the prior state prediction parameters of each sub-observer based on the current observation data corresponding to each sub-observer to obtain the posterior state estimation parameters of each sub-observer includes: Based on the current observation data and observation matrix of each sub-observer, determine the Kalman gain of each sub-observer; Based on the Kalman gain of each sub-observer and the current observation data, the state prediction value is estimated posteriorly to obtain the current posterior estimate value of each sub-observer. The prior error covariance is updated based on the Kalman gain of each sub-observer and the observation matrix to obtain the current posterior error covariance of each sub-observer.

8. The method for estimating vehicle suspension state parameters according to any one of claims 1-4, characterized in that, The interactive state observer is constructed based on all physical characteristic parameters of the vehicle suspension, including: Based on all the physical characteristic parameters of the vehicle suspension, a two-degree-of-freedom model of the suspension is constructed. Determine the time-varying parameters of the target from all physical property parameters; Based on the range of change corresponding to the target time-varying parameter, determine the model transition probability and at least two suspension usage parameters corresponding to the target time-varying parameter; the model transition probability is used to determine the transition probability matrix. Based on the vertical dynamic equations corresponding to the two-degree-of-freedom model of the suspension and at least two suspension operating parameters, at least two sub-observers are constructed; The interactive state observer is constructed based on the at least two sub-observers.

9. The method for estimating vehicle suspension state parameters according to any one of claims 1-4, characterized in that, Based on the vertical dynamic equations corresponding to the two-degree-of-freedom suspension model and the suspension operating parameters, at least two sub-observers are constructed, including: Based on the vertical dynamic equations corresponding to the two-degree-of-freedom suspension model, the state-space equations are determined. Based on the suspension usage parameters and the state space equations, at least two sub-observers are constructed.

10. A vehicle suspension state parameter estimation device, characterized in that, include: A construction module is used to construct an interactive state observer based on all physical characteristic parameters of the vehicle suspension. The interactive state observer includes at least two sub-observers corresponding to each of the physical characteristic parameters. The state prediction module is used to determine the prior state prediction parameters corresponding to each of the sub-observers based on the previous model probability and transition probability matrix corresponding to each of the sub-observers. The observation update module is used to update the prior state prediction parameters of each sub-observer based on the current observation data corresponding to each sub-observer, so as to obtain the posterior state estimation parameters of each sub-observer. The probability update module is used to update the previous model probability of each sub-observer based on the prior state prediction parameters of each sub-observer, the transition probability matrix, and the previous model probability corresponding to the interactive sub-observer, so as to obtain the current model probability. An adaptive determination module is used to adaptively determine the current forgetting factor corresponding to each of the sub-observers based on the current observation data, observation matrix and prior state prediction parameters of each of the sub-observers. The forgetting factor is used to adjust the memory length of each of the sub-observers for historical observation data. The fusion module is used to fuse the current model probabilities, the prior state prediction parameters, and the posterior state estimation parameters corresponding to all sub-observers to obtain the current overall state parameter estimate corresponding to the interactive state observer.

Images