Vehicle state determination method and device, computer equipment and storage medium

Abstract

The invention relates to a vehicle state determination method and device, computer equipment and a storage medium, and the method comprises the steps: constructing a state equation and an observation equation of a vehicle according to a state vector, an input vector and an observation vector; wherein the state vector comprises the speed data of the vehicle and the lateral force of each tire, the input vector comprises the steering angle of each shaft of the vehicle and the longitudinal force of each wheel, and the observation vector comprises the acceleration data of the vehicle and the rotating speed of each wheel; generating a preset number of initial particles; iteratively updating each initial particle according to the state equation and the observation equation to obtain each target particle and the target particle weight of each target particle; and performing weighted summation on each target particle by adopting the target particle weight of each target particle to obtain target vehicle state data of the vehicle. According to the scheme, the advantages of extended Kalman filtering and particle filtering are fused, and the state estimation precision and adaptability in a nonlinear and non-Gaussian environment are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicles, in particular to a vehicle state determination method and device, computer equipment and a storage medium. BACKGROUND

[0002] Vehicle state estimation is the basis for vehicle dynamics control systems to achieve high-performance closed-loop control, and key parameters such as longitudinal speed, lateral speed, yaw rate and tire force need to be obtained in real time and accurately.

[0003] Traditional technologies mostly rely on Kalman filtering (KF), extended Kalman filtering (EKF) or unscented Kalman filtering (UKF), which are all based on the assumption that system noise and observation noise conform to Gaussian distribution. In the scene where noise characteristics are simple, these methods can complete basic state estimation. However, the actual vehicle operating environment is complex, and road excitation, sensor errors and strong nonlinear characteristics of tires will make the system exhibit significant nonlinear and non-Gaussian characteristics, resulting in low accuracy of the determined vehicle state. SUMMARY

[0004] Therefore, it is necessary to provide a vehicle state determination method, device, computer equipment and storage medium to improve the accuracy of the determined vehicle state.

[0005] In a first aspect, the present application provides a vehicle state determination method, comprising:

[0006] According to the state vector, the input vector and the observation vector, the state equation and the observation equation of the vehicle are constructed; wherein the state vector comprises the speed data and the lateral force of each tire of the vehicle, the input vector comprises the steering angle of each axle and the longitudinal force of each wheel of the vehicle, and the observation vector comprises the acceleration data and the rotational speed of each wheel of the vehicle; the state equation is used to describe the state change of the state vector under the influence of the input vector; and the observation equation is used to describe the relationship between the state vector, the input vector and the observation vector;

[0007] A preset number of initial particles are generated; wherein each initial particle represents a candidate value of the initial vehicle state data of the vehicle;

[0008] According to the state equation and the observation equation, each initial particle is iteratively updated to obtain each target particle and the target particle weight of each target particle;

[0009] The target particle weight of each target particle is used to perform weighted summation on each target particle to obtain the target vehicle state data of the vehicle.

[0010] In one of the embodiments, the iterative updating of each initial particle according to the state equation and the observation equation to obtain each target particle and a target particle weight of each target particle comprises:

[0011] For each initial particle, the initial particle is iteratively updated according to the state equation and the observation equation to obtain a posterior particle and a posterior error covariance at the current time;

[0012] A Gaussian proposal distribution of the initial particle is generated according to the posterior particle and the posterior error covariance at the current time;

[0013] The target particle corresponding to the initial particle at the current time is obtained by sampling the Gaussian proposal distribution;

[0014] The target particle weight of the target particle is determined according to the Gaussian proposal distribution, an observation likelihood at the current time and a state transition probability; wherein the observation likelihood at the current time represents a probability of observing the observation vector at the current time when the target particle is the real vehicle state data; and the state transition probability at the current time represents a probability of transferring from the target particle at the previous time to the target particle at the current time.

[0015] In one of the embodiments, the iterative updating of each initial particle according to the state equation and the observation equation to obtain a posterior particle and a posterior error covariance at the current time comprises:

[0016] For each iteration, a prior particle of the current iteration is determined according to the state equation and an intermediate particle;

[0017] A prior error covariance of the current iteration is determined according to the state equation, the intermediate particle and a posterior error covariance corresponding to the intermediate particle;

[0018] A Kalman gain is determined according to the prior error covariance and the observation equation;

[0019] A posterior particle and a posterior error covariance of the current iteration are determined according to the prior particle, the Kalman gain, the observation equation, an observation vector and an input vector at a time corresponding to the current iteration;

[0020] In the case that the time corresponding to the current iteration is the current time, the posterior particle and the posterior error covariance of the current iteration are taken as the posterior particle and the posterior error covariance at the current time;

[0021] Wherein, in the case of the current iteration being the first iteration, the intermediate particle is the initial particle, and the posterior error covariance corresponding to the intermediate particle is an initial value; in the case of the current iteration being a non-first iteration, the intermediate particle is a posterior particle obtained in the last iteration, and the posterior error covariance corresponding to the intermediate particle is a posterior error covariance obtained in the last iteration.

[0022] In one of the embodiments, the target particle weight of the target particle is determined according to the Gaussian proposal distribution, the observation likelihood at the current time and the state transition probability, including:

[0023] The initial particle weight of the target particle is determined according to the Gaussian proposal distribution, the observation likelihood at the current time and the state transition probability;

[0024] The initial particle weight of the target particle is normalized to obtain the target particle weight of the target particle.

[0025] In one of the embodiments, the initial particle weight of the target particle is determined according to the Gaussian proposal distribution, the observation likelihood at the current time and the state transition probability, including:

[0026] The first product of the observation likelihood at the current time and the state transition probability at the current time is determined;

[0027] The initial particle weight of the target particle is determined according to the ratio between the first product and the Gaussian proposal distribution at the current time.

[0028] In one of the embodiments, the state equation and the observation equation of the vehicle are constructed according to the state vector, the input vector and the observation vector, including:

[0029] A low-pass filter is constructed; wherein the transfer function of the low-pass filter is determined according to the reducer gear ratio and the torsional vibration natural frequency corresponding to the wheel;

[0030] The wheel speed of each wheel is input to the low-pass filter to pre-process the wheel speed of each wheel in the observation vector;

[0031] The state equation and the observation equation of the vehicle are constructed according to the state vector, the input vector and the pre-processed observation vector.

[0032] In a second aspect, the application further provides a vehicle state determination device, including:

[0033] The first determining module is configured to construct a state equation and an observation equation of the vehicle according to a state vector, an input vector and an observation vector; the state vector comprises speed data and lateral forces of each tire of the vehicle, the input vector comprises steering angles of each axle and longitudinal forces of each wheel of the vehicle, and the observation vector comprises acceleration data and rotational speeds of each wheel of the vehicle; the state equation is used to describe state changes of the state vector under the influence of the input vector; and the observation equation is used to describe a relationship among the state vector, the input vector and the observation vector;

[0034] The generating module is configured to generate a preset number of initial particles; each initial particle represents a candidate value of initial vehicle state data of the vehicle.

[0035] The second determining module is configured to perform iterative updating on each initial particle according to the state equation and the observation equation, to obtain each target particle and a target particle weight of each target particle.

[0036] The summing module is configured to perform weighted summation on each target particle by using the target particle weight of each target particle, to obtain target vehicle state data of the vehicle.

[0037] In a third aspect, the present application further provides a computer device, comprising a memory and a processor, the memory stores a computer program, and the processor realizes the following steps when executing the computer program:

[0038] The state equation and the observation equation of the vehicle are constructed according to a state vector, an input vector and an observation vector; the state vector comprises speed data and lateral forces of each tire of the vehicle, the input vector comprises steering angles of each axle and longitudinal forces of each wheel of the vehicle, and the observation vector comprises acceleration data and rotational speeds of each wheel of the vehicle; the state equation is used to describe state changes of the state vector under the influence of the input vector; and the observation equation is used to describe a relationship among the state vector, the input vector and the observation vector.

[0039] A preset number of initial particles are generated; each initial particle represents a candidate value of initial vehicle state data of the vehicle.

[0040] Each initial particle is iteratively updated according to the state equation and the observation equation, to obtain each target particle and a target particle weight of each target particle.

[0041] Each target particle is weighted and summed by using the target particle weight of each target particle, to obtain target vehicle state data of the vehicle.

[0042] In a fourth aspect, the present application provides a computer readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the following steps:

[0043] constructing a state equation and an observation equation of the vehicle according to a state vector, an input vector and an observation vector; wherein the state vector comprises speed data and lateral forces of each tire of the vehicle, the input vector comprises steering angles of each axle and longitudinal forces of each wheel of the vehicle, and the observation vector comprises acceleration data and rotational speeds of each wheel of the vehicle; the state equation is used to describe state changes of the state vector under the influence of the input vector; and the observation equation is used to describe the relationship among the state vector, the input vector and the observation vector;

[0044] generating a preset number of initial particles; wherein each initial particle represents a candidate value of initial vehicle state data of the vehicle;

[0045] iteratively updating each initial particle according to the state equation and the observation equation to obtain each target particle and a target particle weight of each target particle;

[0046] performing weighted summation on each target particle by using the target particle weight of each target particle to obtain target vehicle state data of the vehicle.

[0047] In a fifth aspect, the present application provides a computer program product, comprising a computer program which, when executed by a processor, implements the following steps:

[0048] constructing a state equation and an observation equation of the vehicle according to a state vector, an input vector and an observation vector; wherein the state vector comprises speed data and lateral forces of each tire of the vehicle, the input vector comprises steering angles of each axle and longitudinal forces of each wheel of the vehicle, and the observation vector comprises acceleration data and rotational speeds of each wheel of the vehicle; the state equation is used to describe state changes of the state vector under the influence of the input vector; and the observation equation is used to describe the relationship among the state vector, the input vector and the observation vector;

[0049] generating a preset number of initial particles; wherein each initial particle represents a candidate value of initial vehicle state data of the vehicle;

[0050] iteratively updating each initial particle according to the state equation and the observation equation to obtain each target particle and a target particle weight of each target particle;

[0051] performing weighted summation on each target particle by using the target particle weight of each target particle to obtain target vehicle state data of the vehicle.

[0052] The vehicle state determination method, device, computer device and storage medium described above, according to the state vector, the input vector and the observation vector, constructs a state equation and an observation equation of the vehicle; wherein the state vector comprises speed data and each tire lateral force of the vehicle, the input vector comprises each axle steering angle and each wheel longitudinal force of the vehicle, and the observation vector comprises acceleration data and each wheel speed of the vehicle; the state equation is used to describe the state change of the state vector under the influence of the input vector; the observation equation is used to describe the relationship among the state vector, the input vector and the observation vector; a preset number of initial particles are generated; wherein each initial particle represents a candidate value of initial vehicle state data of the vehicle; each initial particle is iteratively updated according to the state equation and the observation equation, to obtain each target particle and a target particle weight of each target particle; each target particle is weighted and summed by using the target particle weight of each target particle, to obtain target vehicle state data of the vehicle. The above scheme combines the advantages of extended Kalman filtering and particle filtering, and improves the state estimation accuracy and adaptability in a nonlinear non-Gaussian environment. BRIEF DESCRIPTION OF DRAWINGS

[0053] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the related art, the drawings needed to be used in the description of the embodiments of the present application or the related art will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and for those skilled in the art, other related drawings can also be obtained without creative labor on the basis of these drawings.

[0054] Figure 1 A flowchart of a vehicle state determination method in an embodiment;

[0055] Figure 2 A flowchart of obtaining each target particle and a target particle weight of each target particle in an embodiment;

[0056] Figure 3 A flowchart of iteratively updating initial particles in an embodiment;

[0057] Figure 4 A vehicle state determination system in an embodiment;

[0058] Figure 5 A structural block diagram of a vehicle state determination device in an embodiment;

[0059] Figure 6 An internal structure diagram of a computer device in an embodiment. DETAILED DESCRIPTION

[0060] In order to make the purposes, technical solutions and advantages of the present application clearer, the present application will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and not intended to limit the present application.

[0061] The vehicle state determination method provided by the embodiments of the present application can be applied to the application scenario of vehicle active suspension control, and is specifically applied to determining the state data of the vehicle to provide high-quality road state information input for the vehicle active suspension control system, so as to improve the stability of the vehicle.

[0062] The method can be executed by a server or a terminal. The server can be a physical server, a server cluster composed of multiple physical servers, or a distributed system, or a cloud server providing cloud computing services. The terminal can be, but is not limited to, various personal computers, notebook computers, smart phones, tablet computers, Internet of Things devices, and portable wearable devices. The Internet of Things device can be a smart speaker, a smart television, a smart air conditioner, a smart vehicle device, a projection device, etc. The portable wearable device can be a smart watch, a smart bracelet, a head-mounted device, etc. The head-mounted device can be a virtual reality (VR) device, an augmented reality (AR) device, smart glasses, etc.

[0063] In an exemplary embodiment, as shown in Figure 1 A vehicle state determination method is provided. The method is described by taking a server as an example and includes the following steps:

[0064] S101, constructing a state equation and an observation equation of the vehicle according to a state vector, an input vector and an observation vector.

[0065] The state vector includes the speed data of the vehicle and the lateral force of each tire, and the speed data of the vehicle includes the longitudinal speed, the lateral speed and the yaw rate of the vehicle. The state vector x is expressed as:

[0066]

[0067] Wherein, u is the longitudinal speed, v is the lateral speed, γ is the yaw rate, F y10 -F yn1 is the lateral force of each tire, and n is the number of tires.

[0068] The input vector includes the steering angle of each axle of the vehicle and the longitudinal force of each wheel. The input vector u is defined as:

[0069]

[0070] Wherein, δ iis the steering angle of the i-th axle, F x10 -F xn1 is the longitudinal force of the i-th tire calculated based on the electric wheel rotation dynamics model.

[0071] The observation vector includes acceleration data of the vehicle and rotational speeds of the wheels. The acceleration data of the vehicle includes longitudinal acceleration, lateral acceleration and yaw rate acceleration of the vehicle. The observation vector y is defined as:

[0072]

[0073] wherein, is the longitudinal acceleration, is the lateral acceleration, is the yaw rate acceleration, - is the rotational speed of the i-th electric wheel. The acceleration data of the vehicle can be measured by an Inertial Navigation System (INS).

[0074] Exemplarily, a state equation and an observation equation of the vehicle can be constructed according to the state vector, the input vector and the observation vector. The state equation is used to describe the state change of the state vector under the influence of the input vector, and the observation equation is used to describe the relationship among the state vector, the input vector and the observation vector. The established state equation is expressed as follows:

[0075]

[0076] wherein, x˙(t) is the time derivative of the state vector x(t), representing the rate of change of the state of the vehicle with time is the state transition function.

[0077] wherein, the state prediction of the tire lateral force adopts a dynamic form of Magic Formula:

[0078]

[0079] wherein, is the i-th tire lateral force of the i-th axle, is is the time derivative of the i-th tire lateral force, i.e. the rate of change of the lateral force; is the tire lateral displacement stiffness, reflecting the change characteristics of the tire lateral force under the action of the lateral displacement; is the initial tire cornering stiffness, representing the strength of the linear relationship between the tire lateral force and the cornering angle under small cornering angle; is the steady-state tire lateral force calculated by Magic Formula, used to describe the lateral force characteristics of the tire under steady-state working condition; is the actual lateral force of the i-th tire of the l-th axle.

[0080]

[0081] wherein X represents a tire parameter, such as tire longitudinal slip ratio, tire side slip angle; B, C, D and E represent stiffness coefficient, shape coefficient, maximum point and curvature coefficient of the corresponding tire force characteristic curve respectively.

[0082] The expression of the observation equation is as follows:

[0083]

[0084] wherein y(t) is an observation vector, is an observation function.

[0085] wherein the relationship between the wheel speed observation value and the state variable is expressed as:

[0086]

[0087] wherein, is the tire longitudinal slip stiffness, r is the effective radius of the tire, is the longitudinal velocity at the center of each tire, is the meaning of "defined as".

[0088] It should be noted that, in order to improve the data quality of the observation vector, the wheel speed in the observation vector can be filtered, and the specific implementation process is as follows:

[0089] A low-pass filter can be constructed; wherein the transfer function of the low-pass filter is determined according to the gear ratio of the corresponding reducer of the wheel and the torsional vibration natural frequency. For example, the transfer function of the low-pass filter is as follows:

[0090]

[0091] wherein wherein is the gear ratio of the reducer, the time constant τw=1 / fcut, the cutoff frequency f cut =f nf / N1, f nf is the torsional vibration natural frequency of the wheel, and N1 is a constant determined by experiment.

[0092] Further, the wheel speed can be input to the low-pass filter to preprocess the wheel speed in the observation vector, for example, to filter out the noise of the wheel speed through the low-pass filter, and to construct the state equation and the observation equation of the vehicle according to the state vector, the input vector and the observation vector after preprocessing. In this way, the quality of the observation data can be improved to facilitate the accuracy of the vehicle state data estimate value.

[0093] Exemplarily, the state equation and the observation equation can be discretized to obtain a discrete state equation and a discrete observation equation.

[0094]

[0095] wherein, is a state vector at time k, is an observation vector at time k, is an input vector at time k, and are process noise and measurement noise respectively and covariance, are defined as Q and R respectively; N is a natural number set, that is, k∈N means that k is a non-negative integer starting from 0, 1, 2, 3, … N, used to represent a discrete time step.

[0096] Since the Extended Kalman Filter - Particle Filter (EKF-PF) algorithm takes the EKF algorithm as the proposal distribution, the EKF algorithm is used to update each time particle. The EKF algorithm needs to linearize the equation, linearize the nonlinear function and around the filter value, and let:

[0097]

[0098]

[0099] The discrete state equation and the discrete observation equation are represented as:

[0100]

[0101]

[0102] wherein, is a state transition matrix, is an input matrix, is an observation matrix, is an input observation matrix.

[0103] S102, generate a preset number of initial particles.

[0104] Exemplarily, the preset number N0 of particles can be generated from the prior probability p(x0) , the weight of all particles is initially set to 1 / N0, and the initial covariance is . The more the total number N0 of particle groups is, the closer the state distribution generated by the PF is to the state posterior distribution. Each initial particle represents a candidate value of the initial vehicle state data of the vehicle.

[0105] S103, iteratively updating each initial particle according to the state equation and the observation equation to obtain each target particle and a target particle weight of each target particle.

[0106] Exemplarily, the "motion trend" of each initial particle under the action of input (such as a steering angle, a wheel longitudinal force) can be simulated based on the state equation to obtain a prior state (i.e., a predicted state at the next time) of the particle by using the state equation of vehicle dynamics.

[0107] Further, the "theoretical observation value" corresponding to each prior particle can be calculated by the observation equation in combination with actual observation data (such as a wheel speed, an acceleration) of a sensor, and compared with a real observation value to adjust the weight of the particle (the higher the weight, the closer the particle to the real state).

[0108] In order to avoid "particle degeneration" (most particle weights tend to 0, only a few effective particles are left), a Gaussian proposal distribution can be generated in combination with EKF, and "target particles" are resampled from the distribution, and the particle weights are updated (the weight sum is 1).

[0109] After iterative updating, a group of new particles obtained are "target particles", and the weight corresponding to each particle reflects its confidence degree to the "real state" at the current time.

[0110] S104, weighted sum of each target particle is performed by using the target particle weight of each target particle to obtain target vehicle state data of the vehicle.

[0111] Exemplarily, the weight of each target particle is a quantitative embodiment of its "credibility", and the higher the weight, the closer the state of the particle to the real value. The state parameters (such as a longitudinal velocity, a lateral velocity, etc.) of all target particles are multiplied by the respective weights and then summed to obtain the final target vehicle state data.

[0112] The vehicle state determination method comprises the following steps: constructing a state equation and an observation equation of the vehicle according to a state vector, an input vector and an observation vector; the state vector comprises speed data and lateral forces of each tire of the vehicle, the input vector comprises steering angles of each axle and longitudinal forces of each wheel of the vehicle, and the observation vector comprises acceleration data and rotational speeds of each wheel of the vehicle; the state equation is used for describing state changes of the state vector under the influence of the input vector; the observation equation is used for describing the relationship among the state vector, the input vector and the observation vector; generating a preset number of initial particles; each initial particle represents a candidate value of initial vehicle state data of the vehicle; iteratively updating each initial particle according to the state equation and the observation equation to obtain target particles and target particle weights of the target particles; and performing weighted summation on the target particles according to the target particle weights of the target particles to obtain target vehicle state data of the vehicle. The above scheme combines the advantages of extended Kalman filtering and particle filtering, and improves the state estimation accuracy and adaptability in a nonlinear non-Gaussian environment.

[0113] In some optional implementations, referring to Figure 2 , Figure 2 A flowchart for obtaining the target particles and the target particle weights of the target particles is provided, and specifically comprises the following steps:

[0114] S201, for each initial particle, iteratively updating the initial particle according to the state equation and the observation equation to obtain a posterior particle and a posterior error covariance at the current time.

[0115] For example, for each initial particle, the state of the current particle can be time-updated according to the state equation to obtain a prior particle; and the prior error covariance is calculated by combining the process noise covariance through the Jacobian matrix (partial derivative with respect to the state vector) of the state equation.

[0116] Further, the theoretical observation value corresponding to the prior particle is calculated by using the observation equation, and compared with the real sensor observation value to obtain the observation residual. The Kalman gain (weighting the weights of the “prediction uncertainty” and the “observation uncertainty”) is calculated by combining the measurement noise covariance R through the Jacobian matrix (partial derivative with respect to the state vector) of the observation equation. The prior particle is corrected by using the Kalman gain to obtain the posterior particle (the optimal state estimation after fusing the kinetic prediction and the observation data). Finally, the posterior error covariance is updated.

[0117] S202, generating a Gaussian proposal distribution of the initial particle according to the posterior particle and the posterior error covariance at the current time.

[0118] For example, the Gaussian distribution of the initial particle can be constructed by taking the posterior particle as the mean value and the posterior error covariance as the covariance matrix, and is expressed as:

[0119]

[0120] wherein, is a Gaussian distribution of the initial particle j, is the jth target particle at time k, is the jth target particle at time k-1, is the observation value at time k, is the covariance matrix of the jth target particle at time k.

[0121] S203, sampling the Gaussian proposal distribution to obtain the target particle corresponding to the initial particle at the current time.

[0122] Exemplarily, the Gaussian proposal distribution can be randomly sampled to obtain the target particle corresponding to the initial particle at the current time. For example, according to the Gaussian probability density with the mean value and the covariance , the Gaussian proposal distribution is randomly sampled to generate a random sample conforming to the distribution, that is, the target particle corresponding to the initial particle at the current time is obtained.

[0123] S204, determining the target particle weight of the target particle according to the Gaussian proposal distribution, the observation likelihood at the current time and the state transition probability.

[0124] wherein, the observation likelihood at the current time represents the probability of observing the observation vector at the current time when the target particle is the real vehicle state data; and the state transition probability at the current time represents the probability of transferring from the target particle at the last time to the target particle at the current time.

[0125] Exemplarily, the target particle weight reflects the degree of fit of the target particle to the "real vehicle state"; the higher the weight, the more likely the target particle represents the real vehicle state. The update of the weight needs to be combined with the "observation data", "state evolution law" and "Gaussian proposal distribution", so that the target particle weight of the target particle can be determined based on the relationship calculation formula, combined with the Gaussian proposal distribution, the observation likelihood at the current time and the state transition probability.

[0126] Exemplarily, the initial particle weight of the target particle can be determined according to the Gaussian proposal distribution, the observation likelihood at the current time and the state transition probability. The calculation formula is as follows:

[0127]

[0128] wherein, is the initial particle weight of the target particle, is proportional to the operation, which means that the weight is determined by the ratio of the numerator to the denominator, and needs to be normalized subsequently, is the observation likelihood at the current time, is a state transition probability of a current time, is a Gaussian proposal distribution.

[0129] That is, the initial particle weight of the target particle can be determined by the following method:

[0130] determining a first product of the observation likelihood of the current time and the state transition probability of the current time ; and determining the initial particle weight of the target particle according to a ratio between the first product and the Gaussian proposal distribution of the current time.

[0131] Exemplarily, the initial particle weight of the target particle can be normalized to obtain a target particle weight of the target particle. Specifically, the normalization can be realized by the following calculation formula:

[0132]

[0133] wherein, is the target particle weight of the target particle, is the initial particle weight of the target particle, and N is the number of target particles and also the number of initial particles.

[0134] Exemplarily, in order to avoid particle degeneration and reduce the operation time on particles with small weights, a resampling step is introduced. For example, the number of effective particles Ne can be calculated, and Ne is compared with a preset threshold to determine whether resampling is needed, wherein the preset threshold can be half of the number of initial particles, and the preset threshold is not specifically limited herein. The calculation formula of Ne is as follows:

[0135]

[0136] If the number of effective particles Ne is less than the preset threshold, resampling is needed to generate a new initial particle group, and each initial particle is iteratively updated again to obtain each target particle and the target particle weight of each target particle. In this way, by screening effective particles with large weights and eliminating invalid particles with weights close to 0, the degeneration problem that most particles lose the representation ability of the true state in particle filtering is avoided. For example, when the number of effective particles is lower than the threshold, resampling retains high-weight particles that fit the true state of the vehicle (such as longitudinal speed and tire lateral force), ensuring that the particle group can always accurately cover the possible range of the true state.

[0137] In the above embodiment, the Gaussian proposal distribution is generated by using the posterior particles and the posterior error covariance, which is used to guide particle sampling, so that particle sampling focuses on the state area with high confidence, avoiding the problem of excessive dispersion or degeneration of particles in particle filtering, ensuring that the particle group always maintains good diversity and effectiveness, so that the particle filtering algorithm can perform state estimation stably for a long time.

[0138] In some optional implementations, referring to Figure 3 , Figure 3 A flowchart for iteratively updating initial particles is provided, and specifically includes the following steps:

[0139] S301, for each iteration, determining the prior particle of the current iteration according to the state equation and the intermediate particle.

[0140] For example, the prior particle of the current iteration can be determined according to the state equation and the intermediate particle based on the following calculation formula.

[0141]

[0142] For example, the prior particle of the current iteration is taken as the k moment. is the jth prior particle at the k moment, is the state transition matrix at the k-1 moment; is the jth posterior particle at the k-1 moment, which is also the intermediate particle; is the input matrix at the k-1 moment, is the input vector at the k-1 moment.

[0143] S302, determining the prior error covariance of the current iteration according to the state equation, the intermediate particle and the posterior error covariance corresponding to the intermediate particle.

[0144] For example, the prior error covariance of the current iteration can be determined according to the state equation, the intermediate particle and the posterior error covariance corresponding to the intermediate particle based on the following calculation formula.

[0145]

[0146] wherein, is the prior error covariance of the jth particle in the current iteration, is the posterior error covariance of the jth particle at the k-1 moment, that is, the posterior error covariance corresponding to the intermediate particle; is the process noise covariance matrix of the jth particle at the k-1 moment.

[0147] S303, determining the Kalman gain according to the prior error covariance and the observation equation.

[0148] For example, the Kalman gain can be determined according to the prior error covariance and the observation equation based on the following calculation formula.

[0149]

[0150] wherein, is the Kalman gain, is the observation matrix transpose, To measure the noise covariance matrix.

[0151] S304. Based on the prior particle, Kalman gain, observation equation, and the observation vector and input vector at the corresponding time of this iteration, determine the posterior particle and posterior error covariance for this iteration.

[0152] For example, the posterior particle and posterior error covariance of the current iteration can be determined based on the following calculation formula, according to the prior particle, Kalman gain, observation equation, and the observation vector and input vector at the corresponding time of the current iteration.

[0153]

[0154] in, Let J be the j-th posterior particle at time k, which is the posterior particle in this iteration. Let be the observation vector at time k, which is the observation vector at the corresponding time in this iteration; Let be the input vector at time k, which is the input vector at the corresponding time of this iteration.

[0155] S305, where the time corresponding to this iteration is the current time, the posterior particle and posterior error covariance of this iteration are used as the posterior particle and posterior error covariance of the current time.

[0156] For example, if the time corresponding to this iteration is the current time, then the posterior particle and posterior error covariance of this iteration are used as the posterior particle and posterior error covariance of the current time.

[0157] In the case of the first iteration, the intermediate particle is the initial particle, and the posterior error covariance corresponding to the intermediate particle is the initial value; in the case of a non-first iteration, the intermediate particle is the posterior particle obtained in the previous iteration, and the posterior error covariance corresponding to the intermediate particle is the posterior error covariance obtained in the previous iteration.

[0158] In the above embodiments, through the iterative update logic of prior particles and posterior particles, each iteration is based on the previous optimization result (intermediate particles and corresponding posterior error covariance), and the particles are gradually refined. Starting from the initial particles, through multiple iterations, the vehicle dynamics model (state equation) and sensor observation information (observation equation) are continuously fused, making the particles closer and closer to the real vehicle state, thus achieving progressively more accurate state estimation.

[0159] In some alternative implementations, see [link to relevant documentation]. Figure 4 , Figure 4 A vehicle state determination system is provided, comprising:

[0160] The sensor and data input module includes:

[0161] Inertial navigation (INS) is used to provide the longitudinal acceleration of a vehicle. Lateral acceleration yaw acceleration It is used to reflect the dynamic motion characteristics of a vehicle.

[0162] Hub motor resolver, used to measure the rotation angle of the motor. The wheel speed ω is obtained after low-pass filtering (LPF). il As one of the direct inputs to the observation system.

[0163] Hub motor voltage and current sensors are used to measure the motor's voltage p. inil The current is used to calculate the motor torque T using a hub motor model. mil Then, the longitudinal force F of the tire is derived through the electric wheel dynamics model. xil .

[0164] Steering wheel / electric wheel angle sensor for measuring wheel steering angle δ i As a quantitative indicator of vehicle control input.

[0165] The vehicle model and state modeling module includes:

[0166] Vehicle dynamics model, used to express the state change rate vector With the core as the basis, combined with the external force matrix Torque matrix T u (V) describes the motion of a vehicle under acceleration and torque, and the formula is: .

[0167] Magic Formula Tire Model, used to model tires using empirical formulas (such as...) , It describes the nonlinear relationship between tire slip angle and lateral force, enabling accurate modeling of tire forces.

[0168] The observation and filtering algorithm module includes:

[0169] Observation vector y: Integrates multi-source observation data, including acceleration, yaw acceleration, wheel speeds, etc., to form a high-dimensional observation input.

[0170] Input vector u: contains wheel angle, tire longitudinal force, motor torque, etc., as a quantitative representation of vehicle control and power input.

[0171] Nonlinear state observation model: through state equations and observation equations A nonlinear mapping from vehicle state to observation output is established.

[0172] Extended Kalman-Particle Filter (EKF-PF): Combining the linear optimization of Extended Kalman Filter and the nonlinear adaptive advantage of Particle Filter, iteratively estimates the state vector and finally outputs high-precision state observations (such as vehicle speed, attitude, tire force, etc.).

[0173] The above embodiment integrates multi-source data such as Inertial Navigation System (INS), wheel motor resolver, voltage and current sensor, and angle sensor, and combines vehicle dynamics model and Magic Formula tire model, constraining vehicle state from both mechanical principle and actual observation, which can achieve high-precision joint estimation of key states such as vehicle longitudinal speed, lateral speed, yaw rate, and tire force, providing accurate state input for vehicle dynamics control and autonomous driving decision.

[0174] The Extended Kalman-Particle Filter (EKF-PF) algorithm combines the ability of Extended Kalman Filter to handle nonlinear systems and the advantage of Particle Filter to handle non-Gaussian distribution, which can effectively deal with the nonlinear and non-Gaussian characteristics of vehicles in complex conditions (such as sudden change of road adhesion coefficient, extreme steering, etc.), ensuring the stability and accuracy of state estimation and improving the robustness of the system.

[0175] Through low-pass filtering (LPF) for pre-processing of wheel speed signals, noise interference is reduced, and resampling and other operations optimize the calculation efficiency of particle filtering. On the premise of ensuring the accuracy of state estimation, it can meet the real-time requirements of vehicle dynamics control system for state information, and can be widely applied to dynamic control scenarios of real vehicles, such as drive control of electric vehicles, ESP (Electronic Stability Program), etc.

[0176] Accurate vehicle state observation is the core foundation of vehicle intelligence and active safety system. The high-precision state data provided by this scheme can enable the vehicle dynamics control system to more accurately identify the motion state of the vehicle and the interaction between the tire and the ground, thereby optimizing the control strategy and improving the vehicle's handling stability, driving safety, and energy utilization efficiency, helping the vehicle to develop towards intelligence and electrification.

[0177] It should be understood that although the steps in the flowcharts involved in the embodiments described above are shown in sequence according to the arrows, the steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified herein, the execution of the steps is not strictly limited in sequence, and the steps can be executed in other orders. Moreover, at least some of the steps in the flowcharts involved in the embodiments described above can include multiple steps or multiple stages, which are not necessarily executed at the same time but can be executed at different times, and the execution of the steps or stages is not necessarily sequential but can be executed in rotation or alternation with at least some of the other steps or the steps or stages in the other steps.

[0178] Based on the same inventive concept, the embodiments of the present application also provide a vehicle state determination apparatus for implementing the vehicle state determination method described above. The implementation scheme for solving the problem provided by the apparatus is similar to the implementation scheme described in the above method, and therefore the specific limitations in one or more vehicle state determination apparatus embodiments provided below can refer to the limitations of the vehicle state determination method described above, which will not be described here again.

[0179] In one exemplary embodiment, as shown in Figure 5 A vehicle state determination apparatus is provided, comprising:

[0180] A first determination module 10 is configured to construct a state equation and an observation equation of a vehicle according to a state vector, an input vector and an observation vector; the state vector includes speed data and lateral forces of each tire of the vehicle, the input vector includes steering angles of each axle and longitudinal forces of each wheel of the vehicle, and the observation vector includes acceleration data and rotational speeds of each wheel of the vehicle; the state equation is used to describe the state change of the state vector under the influence of the input vector; and the observation equation is used to describe the relationship among the state vector, the input vector and the observation vector;

[0181] A generation module 20 is configured to generate a preset number of initial particles; each initial particle represents a candidate value of initial vehicle state data of the vehicle;

[0182] A second determination module 30 is configured to perform iterative updating on each initial particle according to the state equation and the observation equation, to obtain target particles and target particle weights of the target particles;

[0183] A summation module 40 is configured to perform weighted summation on the target particles by using the target particle weights of the target particles, to obtain target vehicle state data of the vehicle.

[0184] In one embodiment, the second determination module 30 specifically includes:

[0185] an updating unit, configured to perform iterative updating on the initial particle according to the state equation and the observation equation to obtain a posterior particle and a posterior error covariance at a current time for each initial particle;

[0186] a generating unit, configured to generate a Gaussian proposal distribution of the initial particle according to the posterior particle and the posterior error covariance at the current time;

[0187] a sampling unit, configured to sample the Gaussian proposal distribution to obtain a target particle corresponding to the initial particle at the current time;

[0188] a determining unit, configured to determine a target particle weight of the target particle according to the Gaussian proposal distribution, an observation likelihood at the current time and a state transition probability; the observation likelihood at the current time represents a probability of observing an observation vector at the current time when the target particle is real vehicle state data; and the state transition probability at the current time represents a probability of transitioning from a target particle at a previous time to the target particle at the current time.

[0189] In one of the embodiments, the updating unit is specifically configured to:

[0190] for each iteration, determine a prior particle of the iteration according to the state equation and an intermediate particle; determine a prior error covariance of the iteration according to the state equation, the intermediate particle and a posterior error covariance corresponding to the intermediate particle; determine a Kalman gain according to the prior error covariance and the observation equation; determine a posterior particle and a posterior error covariance of the iteration according to the prior particle, the Kalman gain, the observation equation, an observation vector at a time corresponding to the iteration and an input vector; and in a case where the time corresponding to the iteration is the current time, take the posterior particle and the posterior error covariance of the iteration as the posterior particle and the posterior error covariance at the current time; wherein in a case where the iteration is a first iteration, the intermediate particle is the initial particle, and the posterior error covariance corresponding to the intermediate particle is an initial value; and in a case where the iteration is a non-first iteration, the intermediate particle is a posterior particle obtained in a previous iteration, and the posterior error covariance corresponding to the intermediate particle is a posterior error covariance obtained in the previous iteration.

[0191] In one of the embodiments, the determining unit specifically includes:

[0192] a determining subunit, configured to determine an initial particle weight of the target particle according to the Gaussian proposal distribution, the observation likelihood at the current time and the state transition probability.

[0193] a normalization subunit configured to normalize the initial particle weight of the target particle to obtain a target particle weight of the target particle.

[0194] In one of the embodiments, the determining subunit is specifically configured to:

[0195] determine a first product of the observation likelihood at the current time and the state transition probability at the current time; and determine the initial particle weight of the target particle according to a ratio between the first product and the Gaussian proposal distribution at the current time.

[0196] In one of the embodiments, the first determining module 10 is specifically configured to:

[0197] construct a low-pass filter, wherein a transfer function of the low-pass filter is determined according to a reducer transmission ratio corresponding to the wheel and a torsional vibration natural frequency; input wheel speeds to the low-pass filter to pre-process the wheel speeds in the observation vector; and construct a state equation and an observation equation of the vehicle according to the state vector, the input vector and the pre-processed observation vector.

[0198] The modules in the vehicle state determining apparatus can be realized by software, hardware or a combination thereof. The modules can be embedded in or independent of a processor in a computer device in a hardware form, or stored in a memory in the computer device in a software form, so as to be called and executed by the processor to perform the operations corresponding to the modules.

[0199] In an exemplary embodiment, a computer device is provided, which can be a server, and an internal structure diagram of the computer device can be as shown in Figure 6 The computer device includes a processor, a memory, an input / output (I / O) interface and a communication interface. The processor, the memory and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. The processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for running the operating system and the computer program in the non-volatile storage medium. The database of the computer device is configured to store a state vector, an input vector and an observation vector, and vehicle state data. The input / output interface of the computer device is configured to exchange information between the processor and external devices. The communication interface of the computer device is configured to communicate with external terminals through a network connection. The computer program is executed by the processor to implement a vehicle state determining method.

[0200] Those skilled in the art can understand that, Figure 6The structure shown in the figure is only a block diagram of part of the structure related to the scheme of the present application, and does not constitute a limitation on the computer device to which the scheme of the present application is applied. The specific computer device can include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0201] In an exemplary embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program, and the processor executing the computer program to implement the steps of the vehicle state determination method described in any of the above embodiments.

[0202] In an embodiment, a computer readable storage medium is provided, storing a computer program, and the computer program being executed by a processor to implement the steps of the vehicle state determination method described in any of the above embodiments.

[0203] In an embodiment, a computer program product is provided, including a computer program, and the computer program being executed by a processor to implement the steps of the vehicle state determination method described in any of the above embodiments.

[0204] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, use and processing of related data need to comply with relevant regulations.

[0205] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a non-volatile computer readable storage medium, and when executed, can include the processes of the above-mentioned embodiment methods. Any reference to memory, database or other medium used in the embodiments provided in the present application can include at least one of non-volatile memory and volatile memory. The non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical storage, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. The volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration but not limitation, the RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM), etc. The database involved in the embodiments provided in the present application can include at least one of a relational database and a non-relational database. The non-relational database can include a distributed database based on a block chain, etc., without being limited thereto. The processor involved in the embodiments provided in the present application can be a general-purpose processor, a central processing unit, a graphics processing unit, a digital signal processor, a programmable logic device, a data processing logic device based on quantum computing, an artificial intelligence (AI) processor, etc., without being limited thereto.

[0206] The technical features of the above embodiments can be combined in any manner. To make the description concise, not all possible combinations of the technical features in the above embodiments are described, but as long as the combinations of the technical features do not exist contradictions, they should be considered as the scope of the present application.

[0207] The above-described embodiments are merely illustrative of several embodiments of the present application, and the description is relatively specific and detailed, but should not be understood as a limitation on the scope of the patent. It should be noted that for those skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are all within the scope of the present application. Therefore, the scope of protection of the present application should be subject to the appended claims.

Claims

1. A vehicle state determination method characterized by, The method comprises: According to the state vector, the input vector and the observation vector, the state equation and the observation equation of the vehicle are constructed; wherein the state vector comprises the speed data and the lateral force of each tire of the vehicle, the input vector comprises the steering angle of each axle and the longitudinal force of each wheel of the vehicle, and the observation vector comprises the acceleration data and the rotational speed of each wheel of the vehicle; the state equation is used to describe the state change of the state vector under the influence of the input vector; and the observation equation is used to describe the relationship among the state vector, the input vector and the observation vector; A preset number of initial particles are generated; wherein each initial particle represents a candidate value of the initial vehicle state data of the vehicle; According to the state equation and the observation equation, each initial particle is iteratively updated to obtain each target particle and the target particle weight of each target particle; The target particle weight of each target particle is used to perform weighted summation on each target particle to obtain the target vehicle state data of the vehicle.

2. The method of claim 1, wherein, According to the state equation and the observation equation, each initial particle is iteratively updated to obtain each target particle and the target particle weight of each target particle, which comprises: For each initial particle, the initial particle is iteratively updated according to the state equation and the observation equation to obtain the posterior particle and the posterior error covariance at the current time; According to the posterior particle and the posterior error covariance at the current time, a Gaussian proposal distribution of the initial particle is generated; The Gaussian proposal distribution is sampled to obtain the target particle corresponding to the initial particle at the current time; According to the Gaussian proposal distribution, the observation likelihood at the current time and the state transition probability, the target particle weight of the target particle is determined; wherein the observation likelihood at the current time represents the probability of observing the observation vector at the current time when the target particle is the real vehicle state data; and the state transition probability at the current time represents the probability of transferring from the target particle at the last time to the target particle at the current time.

3. The method of claim 2, wherein, According to the state equation and the observation equation, each initial particle is iteratively updated to obtain the posterior particle and the posterior error covariance at the current time, which comprises: For each iteration, a prior particle of the current iteration is determined according to the state equation and the intermediate particle; According to the state equation, the intermediate particle and the posterior error covariance corresponding to the intermediate particle, a prior error covariance of the current iteration is determined; According to the prior error covariance and the observation equation, a Kalman gain is determined; According to the prior particle, the Kalman gain, the observation equation, and the observation vector and the input vector at the time corresponding to the current iteration, the posterior particle and the posterior error covariance of the current iteration are determined; In the case that the time corresponding to the current iteration is the current time, the posterior particle and the posterior error covariance of the current iteration are taken as the posterior particle and the posterior error covariance at the current time. Wherein, in the case of the current iteration being the first iteration, the intermediate particle is the initial particle, and the posterior error covariance corresponding to the intermediate particle is an initial value; in the case of the current iteration being a non-first iteration, the intermediate particle is a posterior particle obtained in the last iteration, and the posterior error covariance corresponding to the intermediate particle is a posterior error covariance obtained in the last iteration.

4. The method of claim 2, wherein, According to the Gaussian proposal distribution, the observation likelihood at the current time, and the state transition probability, determining an initial particle weight of the target particle, comprising: According to the Gaussian proposal distribution, the observation likelihood at the current time, and the state transition probability, determining an initial particle weight of the target particle; According to the Gaussian proposal distribution, the observation likelihood at the current time, and the state transition probability, determining an initial particle weight of the target particle.

5. The method of claim 4, wherein, According to the Gaussian proposal distribution, the observation likelihood at the current time, and the state transition probability, determining an initial particle weight of the target particle, comprising: Determining a first product of the observation likelihood at the current time and the state transition probability at the current time; According to a ratio between the first product and the Gaussian proposal distribution at the current time, determining the initial particle weight of the target particle.

6. The method of claim 1, wherein, The state equation and the observation equation of the vehicle are constructed according to the state vector, the input vector and the observation vector, comprising: A low-pass filter is constructed; wherein a transfer function of the low-pass filter is determined according to a reducer transmission ratio and a torsional vibration natural frequency corresponding to the wheel; The wheel speed is input to the low-pass filter to preprocess the wheel speed in the observation vector; The state equation and the observation equation of the vehicle are constructed according to the state vector, the input vector and the preprocessed observation vector.

7. A vehicle state determination device characterized by comprising: The device comprises: A first determining module is configured to construct a state equation and an observation equation of a vehicle according to a state vector, an input vector and an observation vector; wherein the state vector comprises speed data and tire lateral forces of the vehicle, the input vector comprises axle steering angles and wheel longitudinal forces of the vehicle, and the observation vector comprises acceleration data and wheel speeds of the vehicle; the state equation is used to describe state changes of the state vector under the influence of the input vector; and the observation equation is used to describe a relationship among the state vector, the input vector and the observation vector. A generating module is configured to generate a preset number of initial particles; wherein each initial particle represents a candidate value of initial vehicle state data of the vehicle. A second determining module is configured to perform iterative updating on each initial particle according to the state equation and the observation equation, to obtain target particles and target particle weights of the target particles. A summing module is configured to perform weighted summation on the target particles by using the target particle weights of the target particles, to obtain target vehicle state data of the vehicle. 8.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-7. The processor executes the computer program to implement the steps of the method in any one of claims 1 to 6.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method in any one of claims 1 to 6.

10. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the steps of the method in any one of claims 1 to 6.

Images