Front vehicle state estimation method and system based on event triggering and maximum correlation entropy filtering

Abstract

The invention provides a preceding vehicle state estimation method and system based on event triggering and maximum correlation entropy filtering, and relates to the technical field of intelligent driving perception. Comprising the following steps: acquiring front wheel turning angle data and acceleration data of a target vehicle through a vehicle-mounted sensor; dynamically outputting a trigger signal corresponding to the input acceleration data change condition based on a preset event trigger rule according to the input acceleration data change condition; when the trigger signal is 1, transmitting the currently acquired front wheel steering angle data and acceleration data of the target vehicle to a state estimator; when the trigger signal is 0, not transmitting the currently acquired front wheel steering angle data and acceleration data; and the state estimator performs dual-mode calculation on the front wheel steering angle data and the acceleration data based on the trigger signal to obtain a target vehicle state estimation result. Through organic combination of an event triggering mechanism and a maximum correlation entropy extended Kalman filtering algorithm, high-precision and low-communication-load preceding vehicle state estimation is realized.

Description

Technical Field

[0001] The present invention relates to the field of intelligent driving perception technology, and in particular to a method and system for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering. Background Art

[0002] Typical advanced driver assistance systems (ADAS), such as adaptive cruise control and active collision avoidance systems, require real-time perception of the preceding vehicle's motion state to function effectively. Furthermore, the preceding vehicle's motion state is crucial for controlling autonomous vehicles. For example, when a vehicle cuts in from its lane, an autonomous vehicle needs to obtain real-time information about the preceding vehicle's relative distance, speed, and slip angle to determine whether the vehicle can smoothly brake or change lanes to avoid a collision. However, some of the preceding vehicle's state parameters, such as longitudinal speed and slip angle, cannot be directly acquired through on-board sensors. Even if these parameters can be acquired through advanced sensors like lidar, their high cost makes them difficult to implement on a large scale in mass-produced vehicles. Therefore, soft sensing methods, which combine state estimation technology with low-cost on-board sensors, have emerged as a new approach for accurately estimating the preceding vehicle's state.

[0003] Existing research on vehicle state estimation primarily relies on estimation methods based on nonlinear observers. While these methods have proven effective under certain conditions, their estimation accuracy relies heavily on the precise acquisition of vehicle model parameters. Furthermore, methods based on Kalman filters (KFs) have also been widely used for vehicle state estimation. However, traditional KFs are only applicable to linear systems and require the observation system to be linear, making them unsuitable for nonlinear vehicle state estimation systems. Various derivatives of Kalman filters have been proposed to address nonlinear estimation problems, such as the extended Kalman filter (EKF), the unscented Kalman filter (UKF), and the cubic Kalman filter (CKF). However, these methods all assume that the system noise follows a Gaussian distribution. For leading vehicle state estimation systems, it is more realistic to consider the system noise as having a non-Gaussian distribution. However, in real non-Gaussian environments, existing state estimation methods suffer from poor estimation accuracy and robustness.

[0004] At the same time, in a connected environment, as V2V communication has become an indispensable application in intelligent transportation systems, it has become possible for vehicles to exchange inherent parameters such as vehicle size and mass. This also means that the leading vehicle can directly transmit onboard sensor data to the controlled host vehicle through V2V communication for leading vehicle state estimation. In existing research, V2V communication uses a time-triggered communication strategy, that is, communication is carried out at a fixed period, for example, once every 0.1 seconds. In this case, the transmission of the leading vehicle's state information will be discontinuous, that is, during the communication interval, the host vehicle cannot receive the real-time state information of the leading vehicle. The sudden loss of the leading vehicle's state information during the host vehicle's decision-making process is not allowed. In addition, this periodic information transmission method will occupy more communication resources than actually needed.

[0005] The above problems are in urgent need of resolution. Summary of the Invention

[0006] The purpose of the present invention is to provide a method and system for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering. By organically combining the event triggering mechanism with the maximum correlation entropy extended Kalman filter algorithm, the problems of poor accuracy, insufficient robustness and waste of V2V communication resources in the prior art of preceding vehicle state estimation in non-Gaussian noise environments are solved, and high-precision and low-communication-load preceding vehicle state estimation are achieved.

[0007] On the one hand, an embodiment of the present invention provides a method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering, the method comprising: step S1, acquiring the front wheel steering angle data and acceleration data of the target vehicle through the vehicle-mounted sensor; step S2, dynamically outputting a trigger signal corresponding to the change of the input acceleration data based on a preset event trigger rule; step S3, when the trigger signal is 1, transmitting the currently acquired front wheel steering angle data and acceleration data of the target vehicle to the state estimator; when the trigger signal is 0, not transmitting the currently acquired front wheel steering angle data and acceleration data; step S4, the state estimator The meter performs dual-mode calculation on the front wheel angle data and acceleration data based on the trigger signal to obtain a target vehicle state estimation result, including: step S40, when the trigger signal is 1, the maximum correlation entropy extended Kalman filter algorithm is used to calculate the predicted gain matrix for the updated front wheel angle data and acceleration data of the target vehicle at the current moment, and the target vehicle state estimation result is obtained based on the predicted gain matrix; step S41, when the trigger signal is 0, the preset suboptimal gain matrix is used to calculate the target vehicle state estimation result for the front wheel angle data and acceleration data of the target vehicle received at the previous moment.

[0008] Furthermore, the event triggering rules described in step S2 include:

[0009]

[0010] Where, Indicates the measurement value actually used at the current moment; Indicates the measurement value at the previous moment; z k represents the measurement value at the current moment, and ρ represents the threshold for event triggering.

[0011] Furthermore, the step S40 includes: when the trigger signal is 1, receiving the measurement value through the V2V channel, using the prediction gain matrix of the MCCEKF algorithm, initializing the filter, calculating the moment estimation value and the fixed-point iteration algorithm, to obtain the optimal estimation value and the posterior covariance matrix; the step S41 includes: when the trigger signal is 0, using the preset suboptimal gain matrix, filtering based on the measurement value at the previous moment, deriving the posterior estimation error and covariance, and obtaining the optimal estimation value and estimation error covariance when the event trigger is not satisfied.

[0012] Furthermore, the filter initialization process in step S40 includes: initializing the filter state value and its covariance, presetting the bandwidth and iteration threshold; the moment estimation value calculation process in step S40 includes: calculating the prior estimate of the k+1 moment according to the discretized state equation, and calculating the prior estimate of the covariance at the k+1 moment, including:

[0013]

[0014] Where, is the prior estimate of the state variable at time k+1, is the estimated value of the state variable at time k, is the k+1 time covariance prior estimate, is the estimated value of the covariance at time k, F(·) function is the state transfer function, u k is the system input, u k =[δ,a x ] T , δ is the front wheel turning angle, a x is the longitudinal acceleration of the target vehicle, F k is the state transfer Jacobian matrix, Q k is the process noise covariance.

[0015] Furthermore, the fixed-point iterative algorithm in step S40 includes:

[0016] Compute the measure Jacobian matrix:

[0017]

[0018] Where m is the vehicle mass, r is the yaw rate, and v is the vehicle angular velocity.x is the longitudinal speed of the target vehicle, k1 is the lateral stiffness of the front wheel of the target vehicle, k2 is the lateral stiffness of the rear wheel of the target vehicle, a is the distance from the front axle of the target vehicle to the center of mass of the vehicle, b is the distance from the rear axle of the target vehicle to the center of mass of the vehicle, is the estimated value of the state variable at time k+1;

[0019] Fixed-point iteration based on the maximum correlation entropy cost function, including:

[0020] Calculate the predicted gain matrix:

[0021] Iteratively update the target vehicle state estimate based on the prediction gain matrix:

[0022]

[0023] Where z k+1 represents the measurement value at time k+1;

[0024] when Stop the iteration when , and get the posterior estimate As the target vehicle state estimate

[0025] Update the posterior covariance to use as the initial covariance for the next iteration:

[0026]

[0027] Where I is the unit matrix, R k+1 is the measurement noise covariance matrix.

[0028] Furthermore, the maximum relevant entropy cost function is:

[0029]

[0030] Among them, v k+1 is the measurement noise, X k is the state vector of the target vehicle at time k, X k =[β k , v x，k ] T , β k is the front wheel turning angle of the target vehicle at time k, v x，k is the longitudinal velocity of the target vehicle at time k, is the target vehicle state estimation result, Ψ k+1 The covariance matrix of is:

[0031]

[0032] Among them, B K+1，P 、BK+1，r and B K+1 It is obtained by calculating the Cholesky decomposition of the matrix.

[0033] Furthermore, the step S41 also includes, when the trigger signal is 0, using a preset suboptimal gain matrix, filtering based on the measurement value at the previous moment, deriving the posterior estimation error and covariance, and obtaining the optimal estimate value and estimation error covariance when the event trigger is not satisfied, including: using the measurement value at the previous moment to correct and update the prior estimate to obtain the posterior estimation value; obtaining the posterior estimation error based on the posterior estimation value and a preset discrete state equation; obtaining the covariance based on the posterior estimation error calculation; deriving the upper boundary expression of the covariance based on the preset discretized state space expression and event triggering rules; obtaining the optimal gain matrix by deriving the upper boundary expression; and obtaining the optimal estimate value and estimation error covariance based on the optimal gain matrix.

[0034] Furthermore, the method further comprises:

[0035] The method of using the previous moment measurement value to correct and update the prior estimate to obtain the posterior estimate value includes:

[0036]

[0037] Where M k+1 is the prediction gain matrix when the measured change does not meet the event triggering condition, z k+1 is the measurement value at time k+1, Indicates the measurement value actually used at the current moment;

[0038] The obtaining of the a posteriori estimation error based on the a posteriori estimation value and a preset discrete state equation includes:

[0039]

[0040] Where, is the prior estimation error, v k+1 To measure noise;

[0041] The calculating the covariance based on the posterior estimation error includes:

[0042]

[0043] The upper bound expression of the covariance derived based on the preset discretized state space expression and event triggering rules includes:

[0044]

[0045] The obtaining of the optimal gain matrix by deriving the upper boundary expression includes:

[0046]

[0047] Where, represents the covariance of the prior state estimation error at time k+1, represents the measurement covariance predicted at time k+1, σ1 and σ2 are positive scalars.

[0048] Furthermore, obtaining the optimal estimate and the estimation error covariance based on the optimal gain matrix includes: substituting the optimal gain matrix into the posterior estimate calculation formula to obtain the posterior estimate as the optimal estimate; and substituting the optimal gain matrix into the upper boundary expression of the covariance to obtain the estimation error covariance when the event trigger condition is not met.

[0049] In the second aspect, an embodiment of the present invention provides a preceding vehicle state estimation system based on event triggering and maximum correlation entropy filtering, the system comprising: an on-board sensor acquisition module, for acquiring front wheel steering angle data and acceleration data of a target vehicle through an on-board sensor; an event trigger mechanism module, for dynamically outputting a trigger signal corresponding to the change in input acceleration data based on a preset event trigger rule; a sensor data update module, for transmitting the currently acquired front wheel steering angle data and acceleration data of the target vehicle to a state estimator when the trigger signal is 1; and not transmitting the currently acquired front wheel steering angle data when the trigger signal is 0. data and acceleration data; a state estimator, used to perform dual-mode calculation on the front wheel angle data and acceleration data based on the trigger signal to obtain a target vehicle state estimation result, including: when the trigger signal is 1, the maximum correlation entropy extended Kalman filter algorithm is used to calculate the predicted gain matrix for the updated front wheel angle data and acceleration data of the target vehicle at the current moment, and the target vehicle state estimation result is obtained based on the predicted gain matrix; when the trigger signal is 0, the preset suboptimal gain matrix is used to calculate the target vehicle state estimation result for the front wheel angle data and acceleration data of the target vehicle received at the previous moment.

[0050] In a third aspect, an embodiment of the present invention further provides an electronic device comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein when the computer program is executed by the processor, the method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering is implemented.

[0051] In a fourth aspect, an embodiment of the present invention further provides a readable storage medium, which, when the instructions in the storage medium are executed by a processor of an electronic device, enables the electronic device to execute the above-mentioned method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering.

[0052] The beneficial effects of the present invention are as follows: in a networked environment, this method utilizes an event-triggered communication strategy to transmit onboard sensor data, ensuring accurate prediction of the preceding vehicle's motion state even when communication rates are significantly reduced. This communication strategy fully accounts for the increasing number of communicative vehicles, effectively avoiding a dramatic increase in the amount of transmitted data. Furthermore, the maximum correlation entropy criterion is incorporated into the extended Kalman filter algorithm to construct a preceding vehicle state estimator, significantly improving the system's state estimation performance in non-Gaussian noise environments, making it particularly suitable for strongly nonlinear systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The present invention will be further described below with reference to the accompanying drawings and examples.

[0054] Figure 1 This is a flowchart of a preceding vehicle state estimation method based on event triggering and maximum correlation entropy filtering provided in Example 1 of the present invention.

[0055] Figure 2 This is a simplified diagram of a target vehicle dynamics model provided in Example 1 of the present invention.

[0056] Figure 3 This is a schematic diagram of an event-triggered communication mechanism provided by Example 1 of the present invention.

[0057] Figure 4a This is a schematic diagram of the change of lateral acceleration under a double lane-shifting condition generated by simulation software provided in Example 1 of the present invention.

[0058] Figure 4b This is a schematic diagram of the change of the front wheel angle under a double lane change condition generated by simulation software provided in Example 1 of the present invention.

[0059] Figure 4c This is a schematic diagram of changes in longitudinal acceleration under a double lane-shifting condition generated by simulation software and provided in Example 1 of the present invention.

[0060] Figure 5a This is a schematic diagram of the estimation results of the longitudinal speed of the target vehicle by various algorithms under a double lane change condition generated by simulation software provided by Example 1 of the present invention.

[0061] Figure 5b This is a schematic diagram of the estimation errors of the longitudinal speed of a target vehicle by various algorithms under a double lane-changing condition generated by simulation software provided in Example 1 of the present invention.

[0062] Figure 6a This is a schematic diagram of the estimation results of the sideslip angle of the center of mass of a target vehicle by various algorithms under a double lane change condition generated by simulation software provided by Example 1 of the present invention.

[0063] Figure 6b This is a schematic diagram of the error in estimating the sideslip angle of the center of mass of a target vehicle by various algorithms under a double lane-changing condition generated by simulation software provided by Example 1 of the present invention.

[0064] Figure 7 This is a schematic diagram of the number of event triggering during the entire state estimation process of a target vehicle under a double lane change condition generated by simulation software provided by Example 1 of the present invention.

[0065] Figure 8a This is a schematic diagram of lateral acceleration changes under continuous sinusoidal steering conditions generated by simulation software provided in Example 1 of the present invention.

[0066] Figure 8b This is a schematic diagram of the change in front wheel angle under a continuous sinusoidal steering condition generated by simulation software provided in Example 1 of the present invention.

[0067] Figure 8c This is a schematic diagram of longitudinal acceleration changes under continuous sinusoidal steering conditions generated by simulation software and provided in Example 1 of the present invention.

[0068] Figure 9a This is a schematic diagram of the estimation results of the longitudinal speed of a target vehicle by various algorithms under a continuous sinusoidal steering condition generated by simulation software provided by Example 1 of the present invention.

[0069] Figure 9b This is a schematic diagram of the estimation errors of the longitudinal speed of a target vehicle by various algorithms under a continuous sinusoidal steering condition generated by simulation software provided by Example 1 of the present invention.

[0070] Figure 10a This is a schematic diagram of the estimation results of the sideslip angle of the center of mass of a target vehicle by various algorithms under a continuous sinusoidal steering condition generated by simulation software provided by Example 1 of the present invention.

[0071] Figure 10b This is a schematic diagram of the estimation error of the sideslip angle of the center of mass of a target vehicle by various algorithms under a continuous sinusoidal steering condition generated by simulation software provided by Example 1 of the present invention.

[0072] Figure 11 This is a schematic diagram of the number of event triggering during the entire state estimation process of a target vehicle under a continuous sinusoidal steering condition generated by simulation software provided by Example 1 of the present invention.

[0073] Figure 12 This is a structural diagram of a preceding vehicle state estimation system based on event triggering and maximum correlation entropy filtering provided by Example 2 of the present invention.

[0074] Figure 13This is a partial block diagram of an electronic device provided in Example 3 of the present invention. DETAILED DESCRIPTION

[0075] Before discussing the exemplary embodiments in more detail, it should be mentioned that some exemplary embodiments are described as processes or methods depicted as flow charts. Although the flow charts describe the various operations as sequential processes, many of the operations therein can be implemented in parallel, concurrently, or simultaneously. In addition, the order of the various operations can be rearranged. The process can be terminated when its operation is completed, but can also have additional steps not included in the accompanying drawings. The process can correspond to a method, function, procedure, subroutine, subprogram, etc.

[0076] It should be understood that although the terms "first," "second," and the like may be used herein to describe various elements, these elements should not be limited by these terms. These terms are used solely to distinguish one element from another. For example, a first element may be referred to as a second element, and similarly, a second element may be referred to as a first element, without departing from the scope of the exemplary embodiments. The term "and / or" as used herein includes any and all combinations of one or more of the listed associated items.

[0077] The present invention will now be described in detail with reference to the accompanying drawings. This figure is a simplified schematic diagram, which only illustrates the basic structure of the present invention in a schematic manner, and therefore only shows the components related to the present invention.

[0078] Example 1

[0079] For ease of understanding, the following is an overall description of the inventive concept before describing the embodiments of the present invention in detail:

[0080] A method and system for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering is provided. The core principle of this invention is to combine an event-triggered communication mechanism with the Maximum Correlated Entropy Extended Kalman Filter (MCCEKF) algorithm to form a "communication-algorithm" collaborative optimization framework. The event triggering mechanism uses dynamic thresholds to filter valid data, triggering V2V communication only when the state changes significantly, thus addressing the communication resource waste associated with traditional time-triggered strategies. The MCCEKF algorithm uses a maximum correlation entropy cost function to weight high-probability samples, suppressing non-Gaussian noise interference and overcoming the Gaussian noise assumption of the traditional EKF. The vehicle dynamics model uses a bicycle model, establishing longitudinal, lateral, and yaw motion equations. The system dynamics are described using state vectors and input vectors, with lateral acceleration as the measured variable. After the system is discretized, the ET-MCCEKF estimator switches between two modes based on the event triggering result: when the trigger is met, the MCCEKF optimal estimate is obtained through fixed-point iteration. When the trigger is not met, historical data and covariance upper bound control are used to ensure estimation stability, achieving high-precision and low-communication-load preceding vehicle state estimation.

[0081] The specific implementation is as follows:

[0082] like Figure 1 FIG. 1 is a flow chart of a method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering provided by the present invention.

[0083] As an example, the method includes: step S1, obtaining the front wheel angle data and acceleration data of the target vehicle through the on-board sensor; step S2, dynamically outputting the corresponding trigger signal according to the change of the input acceleration data based on the preset event trigger rule; step S3, when the trigger signal is 1, transmitting the currently acquired front wheel angle data and acceleration data of the target vehicle to the state estimator; when the trigger signal is 0, not transmitting the currently acquired front wheel angle data and acceleration data; step S4, the state estimator performs dual-mode calculation on the front wheel angle data and acceleration data based on the trigger signal to obtain the target vehicle state estimation result, including: step S40, when the trigger signal is 1, using the maximum correlation entropy extended Kalman filter algorithm to calculate the prediction gain matrix for the updated front wheel angle data and acceleration data of the target vehicle at the current moment, and obtaining the target vehicle state estimation result based on the prediction gain matrix; step S41, when the trigger signal is 0, using the preset suboptimal gain matrix to calculate the target vehicle state estimation result for the front wheel angle data and acceleration data of the target vehicle received at the previous moment. Among them, the target vehicle is the front vehicle.

[0084] In some feasible implementations, before elaborating on the implementations of the present invention in detail, a front vehicle state estimation system model is first established to provide theoretical support for subsequent calculations.

[0085] First, the front vehicle dynamics modeling: combined with Figure 2 As shown in Figure 2, considering the real-time operation of the state estimator and the accuracy of the model, a bicycle model is used here to describe the dynamic characteristics of the target vehicle in front. The following assumptions are made for the dynamic model of the front vehicle: (1) The influence of the vehicle suspension system is negligible, and the vertical displacement of the vehicle body is assumed to be constant; (2) The influence of the steering system is ignored, and the front wheel angle is assumed to be directly input into the vehicle model; (3) The influence of air and wind resistance during vehicle motion is ignored; (4) During vehicle motion, the pitch motion around the Y axis and the roll motion around the X axis are ignored.

[0086] The bicycle model mainly describes the lateral motion of the vehicle. Considering the longitudinal motion of the vehicle, the longitudinal motion model is combined with the lateral motion model to obtain the vehicle dynamics equation:

[0087]

[0088] Where m is the vehicle mass, δ is the front wheel angle, r is the yaw rate, β is the sideslip angle of the center of mass, v x is the longitudinal speed; a x is the longitudinal acceleration; a y is the lateral acceleration; k1 is the front wheel cornering stiffness; k2 is the rear wheel cornering stiffness; a is the distance from the front axle to the vehicle's center of mass; b is the distance from the rear axle to the vehicle's center of mass.

[0089] The state vector of the system is:

[0090] X=[β,v x ] T ; (3)

[0091] The front wheel angle δ and longitudinal acceleration a x Set as system input:

[0092] u=[δ,a x ] T ; (4)

[0093] Typically, on most commercial vehicles, the front wheel angle δ can be obtained via the on-board controller area network (CAN) bus or steering wheel angle sensor, while the longitudinal acceleration a x It can be obtained through the accelerometer in the vehicle's inertial measurement unit (IMU).

[0094] Secondly, establish the system measurement equation: Before designing the state estimator, define the system's measurement variables. Select the lateral acceleration of the preceding vehicle as the measurement variable, which can be measured by the accelerometer (MEMS) in the IMU:

[0095]

[0096] In summary, the measurement variable of the estimator is expressed as:

[0097] z=[a y ]; (6)

[0098] Combining equations (1) to (4) and (6), the standardized state equation and measurement equation are:

[0099]

[0100] z t =H(X t ,v t ); (8)

[0101] in, and Denote the system state vector and measurement vector, n and m denote the dimension of the vector, respectively; F(·) and H(·) denote the state transfer function and measurement function, respectively; ω t and v t Represent the system process noise and measurement noise respectively; u t Represents the system input vector.

[0102] Again, the system is discretized: In order to estimate the state of the preceding vehicle using discrete sensor data, equations (7) and (8) are discretized. The discretized state space expression obtained by the forward Euler method is:

[0103]

[0104] Among them, X k =[β k , v x,k ] T ;z k =[a y,k ]; k represents the system sampling time; ω k and v k+1 represent process noise and measurement noise respectively, and both satisfy:

[0105]

[0106] Where E(·) represents the expected value; Q k and R k+1 are the covariance matrices of process noise and measurement noise, respectively.

[0107] F k and H k Respectively represent the prior estimation points after linearization processing At the state vector Xk Find the Jacobian matrix of the partial derivatives,

[0108]

[0109] In some feasible implementations, the design process of the front vehicle state estimator is as follows:

[0110] Preferably, the event-triggered communication mechanism process includes: Figure 3 As shown, the event triggering rules described in step S2 include:

[0111]

[0112] Where, Indicates the measurement value actually used at the current moment; Indicates the measurement value at the previous moment; z k Indicates the measurement value at the current moment, z k =a y (k),a y (k) is the lateral acceleration data of the target vehicle at time k, and ρ represents the threshold value for event triggering.

[0113] That is, when λ k = 1, the estimator uses the current measurement value, and the current measurement is transmitted; when λ k = 0, the estimator uses the measurement value at the previous moment, and there is no measurement transmission in the channel.

[0114] Preferably, the step S40 includes: when the trigger signal is 1, receiving the measurement value through the V2V channel, using the prediction gain matrix of the MCCEKF algorithm, initializing the filter, calculating the moment estimation value and the fixed-point iterative algorithm, to obtain the optimal estimation value and the posterior covariance matrix; the step S41 includes: when the trigger signal is 0, using a preset suboptimal gain matrix, filtering based on the measurement value at the previous moment, deriving the posterior estimation error and covariance, and obtaining the optimal estimation value and estimation error covariance when the event trigger is not satisfied.

[0115] Preferably, the filter initialization process in step S40 includes: initializing the filter state value and its covariance, presetting the bandwidth and iteration threshold; the moment estimation value calculation process in step S40 includes: calculating the prior estimate of the k+1 moment according to the discretized state equation, and calculating the prior estimate of the covariance at the k+1 moment, including:

[0116]

[0117] Where, is the prior estimate of the state variable at time k+1, is the estimated value of the state variable at time k, is the k+1 time covariance prior estimate, is the estimated value of the covariance at time k, u k is the system input, u k =[δ,a x ] T , δ is the front wheel turning angle, a x is the longitudinal acceleration of the target vehicle, F k is the state transfer Jacobian matrix, Q k is the process noise covariance.

[0118] Specifically, considering the impact of the event trigger mechanism on measurement update, the state estimator design is divided into two cases. Case 1: When the measurement change satisfies the event trigger mechanism, that is, when λ k+1 =1, the measured value is transmitted to the state estimator via V2V, and the prediction gain matrix K k+1 It is the same as the Maximum Correlated Entropy Extended Kalman Filter (MCCEKF) algorithm. The MCCEKF algorithm design process is as follows:

[0119] (1) Initialize the filter: Initialize the filter state value X0 and its covariance P0. Choose an appropriate kernel bandwidth σ and a small positive number ε.

[0120] (2) Calculate the estimated value at time k+1: The prior estimate of the state and covariance at time k+1 can be expressed as:

[0121]

[0122] Where, is the prior estimate of the state variable at time k+1, is the estimated value of the state variable at time k, is the k+1 time covariance prior estimate, is the estimated value of the covariance at time k, F(·) function is the state transfer function, u k is the system input, u k =[δ,a x ] T , δ is the front wheel turning angle, a x is the longitudinal acceleration of the target vehicle, F k is the state transfer Jacobian matrix, Q k is the process noise covariance.

[0123] Preferably, the fixed-point iterative algorithm in step S40 includes: calculating the measurement Jacobian matrix:

[0124]

[0125] Where m is the vehicle mass, r is the yaw rate, and v is the vehicle angular velocity. x is the longitudinal speed of the target vehicle, k1 is the lateral stiffness of the front wheel of the target vehicle, k2 is the lateral stiffness of the rear wheel of the target vehicle, a is the distance from the front axle of the target vehicle to the center of mass of the vehicle, b is the distance from the rear axle of the target vehicle to the center of mass of the vehicle, is the estimated value of the state variable at time k+1; fixed-point iteration is performed based on the maximum relevant entropy cost function, including:

[0126] Calculate the predicted gain matrix:

[0127] Iteratively update the target vehicle state estimate based on the prediction gain matrix:

[0128]

[0129] Where z k+1 represents the measurement value at time k+1;

[0130] when Stop the iteration when , and get the posterior estimate As the target vehicle state estimate

[0131] Update the posterior covariance to use as the initial covariance for the next iteration:

[0132]

[0133] Where I is the unit matrix, R k+1 is the measurement noise covariance matrix.

[0134] Specifically, after calculating the estimated value at time k+1, the following steps are included:

[0135] (3) Derivation of nonlinear recursive model: From equations (9) and (16), we can obtain:

[0136]

[0137] in,

[0138] Ψ k+1 The covariance matrix of can be expressed as:

[0139]

[0140] Among them, B k+1,p 、B k+1,r and B k+1 It is obtained by calculating the Cholesky decomposition of the matrix.

[0141] Multiply both sides of formula (18) by The nonlinear recursive model is obtained as:

[0142] D k+1 =W k+1 X k+1 +e k+1 ; (19)

[0143] in,

[0144] (4) The fixed-point iterative algorithm process includes: Based on the maximum correlation entropy criterion, the following cost function is defined:

[0145]

[0146] Among them, W k+1,i W k+1 The i-th component of k+1,i Indicates D k+1 The i-th row of ; L = n + m.

[0147] By solving the maximum value solution of the correlation entropy cost function (20), the optimal estimate under the maximum correlation entropy criterion can be obtained as:

[0148]

[0149] Among them, e k+1,i =D k+1,i -W k+1,i X k+1 .

[0150] make You can get:

[0151]

[0152] Due to e k+1,i =D k+1,i -W k+1,i X k+1 , define C k+1,i =G σ (e k+1,i ), you can get:

[0153]

[0154] Where diag(·) represents a diagonal matrix.

[0155] make Get About X k+1 The fixed-point iterative algorithm solves X k+1 The form is:

[0156]

[0157] in,

[0158]

[0159] Then, the posterior covariance matrix is updated as:

[0160]

[0161] During the fixed-point iteration process, when When k=k+1, proceed to the next iteration.

[0162] Preferably, the step S41 also includes, when the trigger signal is 0, using a preset suboptimal gain matrix, filtering based on the measurement value at the previous moment, deriving the posterior estimation error and covariance, and obtaining the optimal estimate value and estimation error covariance when the event trigger is not satisfied, including: using the measurement value at the previous moment to correct and update the prior estimate to obtain the posterior estimation value; obtaining the posterior estimation error based on the posterior estimation value and a preset discrete state equation; obtaining the covariance based on the posterior estimation error calculation; deriving the upper boundary expression of the covariance based on the preset discretized state space expression and the event triggering rule; obtaining the optimal gain matrix by deriving the upper boundary expression; and obtaining the optimal estimate value and estimation error covariance based on the optimal gain matrix.

[0163] Specifically, when the measured change does not satisfy the event trigger mechanism, that is, when λ k+1 =0, the prediction gain matrix is M k+1 At this time, in order to improve the data transmission efficiency, the measurement value at the previous moment is used for filtering. The method of using the measurement value at the previous moment to correct and update the prior estimate to obtain the posterior estimate includes:

[0164]

[0165] Where M k+1 is the prediction gain matrix when the measured change does not meet the event triggering condition, z k+1 is the measurement value at time k+1, Indicates the measurement value actually used at the current moment.

[0166] The posterior estimation error is:

[0167]

[0168] Combined with the discrete state equation (9), the posterior estimation error can be summarized as:

[0169]

[0170] According to the above formula, the covariance It can be expressed as:

[0171]

[0172] in

[0173]

[0174] Based on the system and event triggering mechanism described in formula (9), it is assumed that there is no packet loss and the following conditions are met:

[0175]

[0176] The following inequality can be derived:

[0177]

[0178] Among them, σ1 and σ2 are positive numbers. Substitute the above formula into The expression of , and its upper bound is:

[0179]

[0180] Based on Equations (30) and (36), the upper bound of the covariance It can be expressed as:

[0181]

[0182] right About M k+1 Find the partial derivative and set it to zero, that is The gain matrix is obtained as:

[0183]

[0184] Where, represents the covariance of the prior state estimation error at time k+1, represents the measurement covariance predicted at time k+1, σ1 and σ2 are positive scalars obtained through multiple trial parameters.

[0185] Finally, the gain matrix is substituted into the calculation formula (39) of the upper bound of the covariance to obtain the estimated error covariance when the event triggering condition is not met.

[0186] In some feasible implementations, in order to prove the effectiveness of this method, technicians verified it using more than a dozen simulation analysis methods, as follows: The simulation verification part of the leading vehicle state estimation method designed by the present invention selects two working conditions: double lane change condition and continuous sinusoidal steering condition.

[0187] Under the same conditions, the EKF algorithm and the ET-MCCEKF algorithm proposed in this invention are used to compare the estimation results. In the simulation experiment, the parameters of the preceding vehicle and some parameters related to the ET-MCCEKF algorithm are shown in Table 1.

[0188] Table 1 Parameters of the leading vehicle and state estimator

[0189]

[0190] Double lane shifting condition

[0191] In this case, the state vector is initialized to X0 = [0, 40 / 3.6], the error covariance matrix is initialized to P0 = diag([1, 1]), the road adhesion coefficient is set to 0.85, the simulation time is 15s, and the sampling time is 0.01s.

[0192] The state estimator's measurement signal is lateral acceleration, and the system inputs are the front wheel angle and longitudinal acceleration. The changes in these quantities under the double lane change condition are shown in Figures 4(a)-4(c). The red curve in Figure 4(a) represents the ideal lateral acceleration curve; the green curve represents the signal with non-Gaussian interference added, which is directly used as the measurement signal input for the EKF state estimator; and the blue curve represents the signal with non-Gaussian interference added, which is sent to the ET-MCCEKF state estimator after an event trigger. The curves in Figures 4(b)-(c) represent similar meanings to those in Figure 4(a). The estimation results and estimation errors of the longitudinal speed of the preceding vehicle using various algorithms under the double lane change condition are shown in Figures 5(a)-(b). It can be seen that the estimation results of the ET-MCCEKF method (compared to the traditional filtering method (EKF)) are closer to the ideal value, and the estimation error is smaller than that of the EKF algorithm. Under the double lane change condition, the estimation results and estimation errors of the center of mass slip angle of the leading vehicle by each algorithm are shown in Figure 6(a)-(b). It can be seen that the center of mass slip angle estimation result obtained by this method (ET-MCCEKF) is closer to the ideal value than the traditional filtering method (EKF), and the estimation error obtained is smaller than that of the EKF algorithm. The number of event triggering in the entire state estimation process of the leading vehicle under the double lane change condition is shown in Figure 6(a)-(b). Figure 7 shown.

[0193] Under the continuous sinusoidal steering condition: In this case, the leading vehicle starts at 50 km / h and then decelerates, while the vehicle undergoes a continuous sinusoidal steering test. The system state vector is initialized to X0 = [0, 50 / 3.6], and the error covariance matrix is initialized to P0 = diag([1, 1]). The road adhesion coefficient is set to 0.85, the simulation time is 16 seconds, and the sampling time is 0.01 seconds.

[0194] The state estimator's measurement signal is lateral acceleration, and the system inputs are the front wheel angle and longitudinal acceleration. The changes in these quantities under continuous sinusoidal steering conditions are shown in Figures 8(a)-(c). The red curve in Figure 8(a) represents the ideal lateral acceleration curve; the green curve represents the measurement signal with non-Gaussian interference added, which is input to the EKF state observer; and the blue curve represents the measurement signal input to the ET-MCCEKF state estimator, which is transmitted based on an event-triggered mechanism after adding non-Gaussian interference. The curves in Figures 8(b)-(c) represent similar meanings to those in Figure 8(a). Under continuous sinusoidal steering conditions, the estimation results and estimation errors of the leading vehicle's longitudinal velocity by each algorithm are shown in Figures 9(a)-(b). It can be seen that under continuous sinusoidal steering conditions, the proposed method (ET-MCCEKF) achieves an estimation result closer to the ideal value and a smaller estimation error than the traditional filtering method (EKF). Under continuous sinusoidal steering conditions, the estimation results and estimation errors of the center of mass slip angle of the leading vehicle by each algorithm are shown in Figure 10(a)-(b). It can be seen that compared with the traditional filtering method (EKF), the center of mass slip angle estimation result obtained by this method (ET-MCCEKF) is closer to the ideal value, and the estimation error obtained is smaller than that of the EKF algorithm. The number of event triggers in the entire state estimation process of the leading vehicle under continuous sinusoidal steering conditions is shown in Figure 10(a)-(b). Figure 11 shown.

[0195] Based on the above simulation results, the ET-MCCEKF algorithm proposed in this invention can effectively suppress non-Gaussian noise interference under double lane change and continuous sinusoidal steering conditions, compared with the traditional EKF algorithm, and improve the accuracy and robustness of the leading vehicle state estimation. At the same time, it reduces the number of communications by 35%-40% through the event trigger mechanism, achieving a balance between high-precision estimation and low communication load, and providing an efficient and reliable solution for the leading vehicle state estimation of intelligent connected vehicles.

[0196] Example 2

[0197] See also Figure 12 , this embodiment provides a schematic diagram of the structure of a preceding vehicle state estimation system based on event triggering and maximum correlation entropy filtering.

[0198] As an example, the system adopts the preceding vehicle state estimation method based on event triggering and maximum correlation entropy filtering described in Example 1, and the system includes:

[0199] The vehicle-mounted sensor acquisition module 1 is used to acquire the front wheel angle data and acceleration data of the target vehicle through the vehicle-mounted sensor.

[0200] The event trigger mechanism module 2 is used to dynamically output a corresponding trigger signal according to the change of the input acceleration data based on the preset event trigger rules.

[0201] The sensor data update module 3 is used to transmit the currently acquired front wheel angle data and acceleration data of the target vehicle to the state estimator when the trigger signal is 1; and not transmit the currently acquired front wheel angle data and acceleration data when the trigger signal is 0.

[0202] The state estimator 4 is used to perform dual-mode calculation on the front wheel angle data and acceleration data based on the trigger signal to obtain the target vehicle state estimation result, including: when the trigger signal is 1, the maximum correlation entropy extended Kalman filter algorithm is used to calculate the prediction gain matrix for the updated front wheel angle data and acceleration data of the target vehicle at the current moment, and the target vehicle state estimation result is obtained based on the prediction gain matrix; when the trigger signal is 0, the preset suboptimal gain matrix is used to calculate the target vehicle state estimation result for the front wheel angle data and acceleration data of the target vehicle received at the previous moment.

[0203] It is not difficult to find that this embodiment is a system embodiment corresponding to the first embodiment, and this embodiment can be implemented in conjunction with the first embodiment. The relevant technical details mentioned in the first embodiment are still valid in this embodiment, and to reduce repetition, they are not repeated here. Accordingly, the relevant technical details mentioned in this embodiment can also be applied to the first embodiment.

[0204] It is worth noting that all modules involved in this embodiment are logical units. In actual applications, a logical unit can be a physical unit, a part of a physical unit, or a combination of multiple physical units. Furthermore, to highlight the innovations of this invention, this embodiment does not include units that are not closely related to solving the technical problems proposed by this invention. However, this does not mean that other units do not exist in this embodiment.

[0205] Example 3

[0206] See also Figure 13 An embodiment of the present invention also provides an electronic device, comprising: a memory and a processor; the memory stores at least one program instruction; the processor loads and executes the at least one program instruction to implement the preceding vehicle state estimation method based on event triggering and maximum correlation entropy filtering provided in Example 1.

[0207] The memory 702 and processor 701 are connected using a bus. The bus can include any number of interconnected buses and bridges, connecting various circuits of one or more processors 701 and memory 702. The bus can also connect various other circuits such as peripheral devices, voltage regulators, and power management circuits. These are all well known in the art and, therefore, are not described further herein. The bus interface provides an interface between the bus and the transceiver. The transceiver can be a single component or multiple components, such as multiple receivers and transmitters, providing a unit for communicating with various other devices over a transmission medium. Data processed by the processor 701 is transmitted over a wireless medium via an antenna. Furthermore, the antenna receives data and transmits it to the processor 701.

[0208] The processor 701 is responsible for managing the bus and general processing, and can also provide various functions, including timing, peripheral interfaces, voltage regulation, power management, and other control functions. The memory 702 can be used to store data used by the processor 701 when performing operations.

[0209] Example 4

[0210] An embodiment of the present invention further provides a storage medium storing a method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering. When executed by a processor, the program for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering implements the steps of the method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering as described above. Because this storage medium utilizes all the technical solutions of all the aforementioned embodiments, it at least possesses all the beneficial effects brought about by the technical solutions of the aforementioned embodiments, which will not be detailed here.

[0211] The above is only an embodiment of the present invention. Common knowledge such as the known specific structures and characteristics in the scheme is not described in detail here. Ordinary technicians in the field are aware of all common technical knowledge in the technical field of the invention before the application date or priority date, can obtain all existing technologies in the field, and have the ability to apply conventional experimental means before that date. Ordinary technicians in the field can improve and implement this scheme in combination with their own abilities under the inspiration given by this application. Some typical known structures or known methods should not become obstacles for ordinary technicians in the field to implement this application. It should be pointed out that for those skilled in the art, without departing from the structure of the present invention, several variations and improvements can be made, which should also be regarded as the scope of protection of the present invention. These will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the specification can be used to interpret the content of the claims.

Claims

1. A method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering, characterized in that: The method comprises: Step S1, obtaining front wheel angle data and acceleration data of the target vehicle through the vehicle-mounted sensor; Step S2: dynamically outputting a trigger signal corresponding to the change in the input acceleration data based on a preset event trigger rule; Step S3: when the trigger signal is 1, the currently acquired front wheel steering angle data and acceleration data of the target vehicle are transmitted to the state estimator; when the trigger signal is 0, the currently acquired front wheel steering angle data and acceleration data are not transmitted; Step S4: The state estimator performs dual-mode calculation on the front wheel angle data and the acceleration data based on the trigger signal to obtain a target vehicle state estimation result, including: Step S40: When the trigger signal is 1, a maximum correlation entropy extended Kalman filter algorithm is used to calculate a prediction gain matrix based on the updated front wheel angle data and acceleration data of the target vehicle at the current moment, and a target vehicle state estimation result is obtained based on the prediction gain matrix; Step S41: When the trigger signal is 0, a preset suboptimal gain matrix is used to calculate the target vehicle state estimation result based on the front wheel angle data and acceleration data of the target vehicle received at the last moment.

2. The method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering according to claim 1, characterized in that: The event triggering rules described in step S2 include: Where, Indicates the measurement value actually used at the current moment; Indicates the measurement value at the previous moment; z k represents the measurement value at the current moment, and ρ represents the threshold for event triggering.

3. The method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering according to claim 1, characterized in that: The step S40 includes: when the trigger signal is 1, receiving the measurement value through the V2V channel, using the prediction gain matrix of the MCCEKF algorithm, initializing the filter, calculating the time estimate value and the fixed-point iterative algorithm, to obtain the optimal estimate value and the posterior covariance matrix; The step S41 includes: when the trigger signal is 0, using a preset suboptimal gain matrix, filtering based on the measurement value at the previous moment, deriving the posterior estimation error and covariance, and obtaining the optimal estimation value and estimation error covariance when the event trigger is not satisfied.

4. The method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering according to claim 3, characterized in that: The filter initialization process in step S40 includes: Initialize the filter state value and its covariance, select the kernel bandwidth and iteration threshold; The moment estimation value calculation process in step S40 includes: Calculate the prior estimate of the k+1 moment based on the discretized state equation, and calculate the prior estimate of the covariance at the k+1 moment, including: Where, is the prior estimate of the state variable at time k+1, is the estimated value of the state variable at time k, is the k+1 time covariance prior estimate, is the estimated value of the covariance at time k, F(·) function is the state transfer function, u k is the system input, u k =[δ,a x ] T , δ is the front wheel turning angle, a x is the longitudinal acceleration of the target vehicle, F k is the state transfer Jacobian matrix, Q k is the process noise covariance.

5. The method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering according to claim 4, characterized in that: The fixed-point iterative algorithm in step S40 includes: Compute the measure Jacobian matrix: Where m is the vehicle mass, r is the yaw rate, and v is the vehicle angular velocity. x is the longitudinal speed of the target vehicle, k1 is the lateral stiffness of the front wheel of the target vehicle, k2 is the lateral stiffness of the rear wheel of the target vehicle, a is the distance from the front axle of the target vehicle to the center of mass of the vehicle, b is the distance from the rear axle of the target vehicle to the center of mass of the vehicle, is the estimated value of the state variable at time k+1; Fixed-point iteration based on the maximum correlation entropy cost function, including: Calculate the predicted gain matrix: Iteratively update the target vehicle state estimate based on the prediction gain matrix: Where z k+1 represents the measurement value at time k+1; when Stop the iteration when , and get the posterior estimate As the target vehicle state estimate Update the posterior covariance to use as the initial covariance for the next iteration: Where I is the unit matrix, R k+1 is the measurement noise covariance matrix.

6. The method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering according to claim 5, characterized in that: The maximum relevant entropy cost function is: Among them, v k+1 is the measurement noise, X k is the state vector of the target vehicle at time k, X k =[β k , v x，k ] T , β k is the front wheel turning angle of the target vehicle at time k, v x，k is the longitudinal velocity of the target vehicle at time k, is the target vehicle state estimation result, Ψ k+1 The covariance matrix of is: Among them, B K+1，P 、B K+1，r and B K+1 It is obtained by calculating the Cholesky decomposition of the matrix.

7. The method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering according to claim 5, characterized in that: The step S41 further includes, when the trigger signal is 0, using a preset suboptimal gain matrix, filtering based on the measurement value at the previous moment, deriving the posterior estimation error and covariance, and obtaining the optimal estimation value and estimation error covariance when the event trigger is not satisfied, including: Use the measurement value at the previous moment to correct and update the prior estimate to obtain the posterior estimate; Obtaining a posteriori estimation error based on the posteriori estimation value and a preset discrete state equation; Calculating a covariance based on the posterior estimation error; The upper bound expression of the covariance is derived based on the preset discretized state space expression and event triggering rules; Obtaining an optimal gain matrix by deriving the upper boundary expression; An optimal estimation value and an estimation error covariance are obtained based on the optimal gain matrix.

8. The method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering according to claim 7, characterized in that: The method further comprises: The method of using the previous moment measurement value to correct and update the prior estimate to obtain the posterior estimate value includes: Where M k+1 is the prediction gain matrix when the measured change does not meet the event triggering condition, z k+1 is the measurement value at time k+1, Indicates the measurement value actually used at the current moment; The obtaining of the a posteriori estimation error based on the a posteriori estimation value and a preset discrete state equation includes: Where, is the prior estimation error, v k+1 To measure noise; The calculating the covariance based on the posterior estimation error includes: The upper bound expression of the covariance derived based on the preset discretized state space expression and event triggering rules includes: The obtaining of the optimal gain matrix by deriving the upper boundary expression includes: Where, represents the covariance of the prior state estimation error at time k+1, represents the measurement covariance predicted at time k+1, σ1 and σ2 are positive scalars.

9. The method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering according to claim 8, characterized in that: Obtaining an optimal estimated value and an estimated error covariance based on the optimal gain matrix includes: Substitute the optimal gain matrix into the posterior estimation value calculation formula to obtain the posterior estimation value as the optimal estimation value; Substituting the optimal gain matrix into the upper bound expression of the covariance, the estimated error covariance is obtained when the event triggering condition is not met.

10. A system for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering, wherein the system is implemented using the method for estimating the state of a preceding vehicle based on event triggering and maximum correlation entropy filtering according to any one of claims 1 to 9, and wherein: The system comprises: A vehicle-mounted sensor acquisition module, configured to acquire front wheel turning angle data and acceleration data of a target vehicle through a vehicle-mounted sensor, wherein the acceleration data includes longitudinal acceleration data and lateral acceleration data; The event trigger mechanism module is used to dynamically output a trigger signal corresponding to the change of the input acceleration data based on the preset event trigger rules; a sensor data updating module, configured to transmit the currently acquired front wheel steering angle data and acceleration data of the target vehicle to the state estimator when the trigger signal is 1; and not transmit the currently acquired front wheel steering angle data and acceleration data when the trigger signal is 0; A state estimator is used to perform dual-mode calculations on the front wheel angle data and acceleration data based on the trigger signal to obtain a target vehicle state estimation result, including: when the trigger signal is 1, using the maximum correlation entropy extended Kalman filter algorithm to calculate the predicted gain matrix for the updated front wheel angle data and acceleration data of the target vehicle at the current moment, and obtaining the target vehicle state estimation result based on the predicted gain matrix; when the trigger signal is 0, using a preset suboptimal gain matrix to calculate the target vehicle state estimation result for the front wheel angle data and acceleration data of the target vehicle received at the previous moment.

Images