Vehicle state estimation system and method based on space-ground integrated network

Abstract

The invention discloses a vehicle state estimation system and method based on a space-ground integrated network. The system comprises a vehicle-mounted LEO signal receiver, a data processor, a vehicle-mounted sensor, a vehicle state estimation system and a stability control system. The vehicle-mounted LEO signal receiver communicates with a low-orbit satellite and receives satellite data in real time so as to obtain information of a vehicle in real time; the data processor performs data cleaning on the data received by the vehicle-mounted LEO signal receiver; the vehicle-mounted sensor unit measures the angular velocity and the linear acceleration of the vehicle in real time; the vehicle state estimation system is fused with vehicle-mounted sensor information and satellite information to estimate the vehicle side slip angle; the stability control system comprises a stability judgment basis module and an active safety control module, stability judgment is conducted according to the side slip angle output by the vehicle state estimation system, and active safety control is started according to the stability judgment result.

Description

Technical Field

[0001] The present invention relates to the field of intelligent vehicle control technology, and in particular to a vehicle state estimation system and method based on a ground-space integrated network. Background Art

[0002] With the rapid development of artificial intelligence, autonomous driving technology has become a hot topic in current research. Autonomous commercial vehicles are also gradually entering the market. Due to their high load capacity, reinforced structural design, and powerful powertrains, these vehicles require more reliable active safety control systems. Vehicle stability control is a core module in active safety technology research. Its primary control objective is to accurately estimate key information such as the vehicle's center of mass, sideslip angle, yaw angle, and angular velocity.

[0003] Existing estimation methods can be roughly divided into two categories: model-based methods and neural network-based methods. Model-based methods typically combine vehicle dynamics and tire models with specialized state observers to estimate the vehicle's sideslip angle. However, the nonlinear characteristics of tires and the randomness of observation noise can reduce the accuracy of the results. Neural network-based methods require a large amount of training data as a basis, thereby improving robustness to changing driving conditions and different sensor configurations.

[0004] Currently, there are also methods to estimate the vehicle's center of mass sideslip angle by combining the vehicle kinematic model with the global positioning system (GPS), inertial measurement unit (IMU) and other kinematics-related sensors. However, the GPS measurement value has a low update frequency, and measurement anomalies and signal synchronization also bring huge challenges to accurately estimating the center of mass sideslip angle. At the same time, measurements based solely on on-board sensors will rely too much on the accuracy of vehicle dynamics modeling. When the vehicle is under extreme conditions, it will bring more challenges to the dynamics modeling, thereby reducing the accuracy of state estimation.

[0005] In summary, due to the shortcomings of existing vehicle state estimation methods, a more accurate vehicle state estimation method is urgently needed. Low-Earth Orbit satellite signal strength has increased by 100-1000 times, significantly enhancing its ability to penetrate urban buildings. It can still provide stable positioning in "urban canyons" where GPS fails, while also significantly increasing the frequency of data updates. This provides a new solution for vehicle state estimation. Summary of the Invention

[0006] In order to address the deficiencies in the prior art, the present application proposes a vehicle state estimation system and method based on a space-ground integrated network. The system integrates low-orbit satellite data in the space-ground integrated network with on-board sensor data. The low-orbit satellite constellation, with the physical advantage of low-Earth orbit (300-600km), can provide Gbps-level data throughput and minute-level update frequency, fundamentally avoiding the problem of a surge in on-board controller load caused by long-cycle data in traditional navigation systems. At the same time, due to the large amount of data, redundant data is also brought to the on-board processor, resulting in an excessive computational load on the processor. Therefore, the data transmitted by the low-orbit satellite is first cleaned, and then a tightly coupled observation matrix input improved cubature Kalman filter is formed with the on-board sensor to achieve accurate vehicle state estimation.

[0007] The technical solutions adopted in the present invention are as follows:

[0008] A vehicle state estimation system based on a ground-space integrated network, the system comprising: a vehicle-mounted LEO signal receiver, a data processor, a vehicle-mounted sensor, a vehicle state estimation system, and a stability control system;

[0009] The vehicle-mounted LEO signal receiver communicates with a low-orbit satellite and receives satellite data in real time to obtain vehicle information in real time;

[0010] The data processor cleans the data received by the vehicle-mounted LEO signal receiver;

[0011] The on-board sensor unit measures the angular velocity and linear acceleration of the vehicle in real time;

[0012] The vehicle state estimation system estimates the vehicle's center of mass sideslip angle by fusing onboard sensor information and satellite information;

[0013] The stability control system includes a stability judgment module and an active safety control module. The stability judgment module performs stability judgment based on the center of mass sideslip angle output by the vehicle state estimation system, and the active safety control module activates active safety control based on the stability judgment result.

[0014] Furthermore, in the data processor, the speed in the NE global coordinate is transformed to obtain the two-dimensional plane vehicle speed v s , recorded as:

[0015]

[0016] Among them, v x represents the longitudinal velocity of the vehicle, v y Represents the lateral velocity of the vehicle.

[0017] Furthermore, the vehicle-mounted sensor unit uses a vehicle-mounted IMU to measure the angular velocity and linear acceleration of the vehicle in real time.

[0018] Furthermore, in the vehicle state estimation system, the maximum correlation entropy square root cubic Kalman filter algorithm is used for state estimation.

[0019] Furthermore, the maximum correlation entropy square root cubature Kalman filter method introduces the maximum correlation entropy MCC criterion on the basis of SCKF.

[0020] Furthermore, the specific derivation process of MCSCKF is as follows:

[0021] The nonlinear recursive model is established as:

[0022]

[0023] Among them, x(k) is the state vector, which is recorded as x(k)=[v x v y r ψ] T ; z(k) is the measurement vector, recorded as z(k)=[rv e,LEO v n,LEO ]; is the state vector at time k-1; h(.) is the measurement function; φ(k) is the process noise nonlinear recursive matrix, denoted as w(k) is the process noise;

[0024] The best estimate of x(k) based on MCC is obtained as follows,

[0025]

[0026] Among them, n is the dimension of the state vector, m is the dimension of the measurement vector, e(k) is the model prediction error, and e i (k) is the i-th component of e(k), G σ is the Gaussian kernel function;

[0027] By making the first-order derivative of the above equation zero, we can get the optimal solution of the equation.

[0028]

[0029] Among them, J MCC is the objective function; if we define C i (k) = G σ (e i (k)),

[0030]

[0031] In practice, the true state x(k) is usually unknown. Therefore, the prior measurement noise variance is written as the measurement noise is:

[0032]

[0033] The covariance matrix can be obtained as follows:

[0034]

[0035] Among them, M r (k) is the weighted square root matrix measuring the prediction error.

[0036] Furthermore, the sideslip angle of the center of mass is as follows:

[0037] Furthermore, in the stability judgment module, the side slip angle β output by the vehicle state estimation system is calculated based on the phase plane. As a basis for stability judgment, when the stability boundary condition of the vehicle yaw-side phase plane is met When , the active safety control unit is not activated, otherwise, the active safety control unit needs to be activated. β is the center of mass side slip angle, Its derivative represents the rate of change of the sideslip angle, k1 and k2 are coefficients related to the road adhesion coefficient and the front wheel angle, and k1 is a dynamic response time constant used to balance the sideslip angle β and its rate of change. It reflects the vehicle system’s sensitivity to the rate of change of the slip angle. k2 is the boundary amplitude of the stable region, representing the maximum equivalent slip energy that the vehicle can withstand.

[0038] Furthermore, the vehicle state estimation system fuses low-orbit satellite data with vehicle-mounted sensor data to perform state estimation as follows:

[0039]

[0040] Among them, I z is the vehicle's moment of inertia, ψ represents the vehicle's yaw angle, r represents the vehicle's yaw angular velocity, and F yf , F yr represents the lateral force of the front and rear tires, l f 、l r Represents the stiffness coefficient of the front and rear tires, are the vehicle's longitudinal velocity v x , the vehicle's lateral velocity v y The first derivative of .

[0041] A vehicle state estimation method based on a space-ground integrated network utilizes a vehicle state estimation system based on a space-ground integrated network to perform vehicle state estimation. The beneficial effects of the present invention are as follows:

[0042] 1. The core role of the data processor in cleaning the received low-orbit satellite data is to convert the original observation values ​​into high-precision, low-noise, and time- and space-consistent reliable inputs, eliminating the multipath effect and signal jitter caused by high-speed satellite motion. After cleaning, the data volume is reduced by 60%-70%, thereby improving the iteration speed of the Kalman filter.

[0043] 2. The state estimation module integrates low-orbit satellite data with onboard sensor data. Low-orbit satellites orbit at an altitude of only 1 / 10 to 1 / 60 that of traditional GNSS satellites, offering 100–1000 times greater signal strength and greater penetration into urban buildings. Even in "urban canyons" where traditional GNSS fails, low-orbit satellites can still provide stable signals, increasing positioning availability from 70% to 98.5%. LEO satellites, acting as "sky reference stations," broadcast centimeter-level orbit and clock corrections. Combined with their high-speed motion, they can achieve real-time centimeter-level positioning over the entire area by fixing carrier phase ambiguities within seconds. Combining these advantages, integrating LEO satellite data improves the robustness and accuracy of the vehicle state estimation system.

[0044] 3. The maximum entropy square root cubic Kalman filter is used to effectively handle the uncertainty and noise problems in the high-dimensional state space through nonlinear transformation and statistical inference, thereby obtaining a high-precision vehicle state β. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Figure 1 are geodetic coordinates.

[0046] Figure 2 are the vehicle's travel coordinates.

[0047] Figure 3 This is the block diagram of the autonomous driving vehicle state estimation system. DETAILED DESCRIPTION

[0048] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0049] A vehicle state estimation system based on a ground-ground integrated network includes an on-board LEO signal receiver, a data processor, on-board sensors, a vehicle state estimation system, and a stability control system. The connections and functions of each component are as follows:

[0050] (1) On-board LEO signal receiver: The on-board LEO satellite signal receiver is installed in the center of the vehicle roof. The on-board LEO signal receiver communicates with the LEO satellite, receives satellite data in real time, and obtains vehicle information in real time, providing important decision support for the on-board control system. This enables full-area centimeter-level positioning and continuous speed measurement, ensuring high-precision synchronous updates of vehicle position and speed in complex environments.

[0051] (2) Data processor: The data processor cleans the data received by the vehicle-mounted LEO signal receiver. The data cleaning refers to the adaptive down-conversion pre-processing of the received LEO data and outputs the processed data. LEO The two-dimensional plane speed v s , ensuring that the onboard controller can still operate smoothly under high load conditions, meet the vehicle status recognition, and achieve stable active safety control. The details are as follows:

[0052] like Figure 1 As shown, the velocity component v in the NE global coordinate is e,LEO and v n,LEO They are as follows:

[0053]

[0054] in, represents the heading angle, β represents the sideslip angle of the vehicle's center of mass, and ψ is the vehicle's yaw angle. Figure 2 Schematic diagram representing this.

[0055] The two-dimensional plane speed is recorded as:

[0056]

[0057] Among them, v x represents the longitudinal velocity of the vehicle, v y Represents the lateral velocity of the vehicle. Both vehicle speed and heading angle are measured using low-orbit satellites and a line of sight between the vehicle and the satellite.

[0058] (3) On-board sensor unit: The on-board IMU (Inertial Measurement Unit) is the core sensor of the vehicle perception system. Its core value lies in real-time measurement of the vehicle's angular velocity and linear acceleration, providing underlying data support for positioning, attitude control and safety systems. The lateral acceleration and longitudinal acceleration are defined as: a x ,a y .

[0059] (4) Vehicle state estimation system: In the vehicle state estimation system, by fusing the vehicle sensor information a x ,a y Information about low-orbit satellites e,LEO ,v n,LEOThe vehicle's center of mass sideslip angle is estimated. This section uses the Maximum Correlated Entropy Square Root Cubic Kalman Filter (MCCSCKF) algorithm as an example. The center of mass sideslip angle is calculated as follows:

[0060]

[0061] More specifically, the vehicle state estimation system fuses low-orbit satellite data (LEO) with onboard sensor data (IMU) to perform state estimation as follows:

[0062]

[0063] Among them, I z is the vehicle's moment of inertia, ψ represents the vehicle's yaw angle, r represents the vehicle's yaw angular velocity, and F yf , F yr represents the lateral force of the front and rear tires, l f 、l r Represents the stiffness coefficient of the front and rear tires, are the vehicle's longitudinal velocity v x , the vehicle's lateral velocity v y The first derivative of .

[0064] (5) Stability control system: The stability control system includes a stability judgment module and an active safety control module. First, the system estimates the side slip angle β output by the vehicle state, and then calculates the side slip angle β of the center of mass according to the phase plane. As a basis for stability judgment, when When the vehicle is in the correct position, the active safety control unit is not activated. Otherwise, the active safety control unit is activated, such as direct yaw moment control, to prevent the vehicle from rolling over. k1 and k2 are coefficients related to the road adhesion coefficient and the front wheel angle, and can be determined through extensive experimental data in the early stages.

[0065] More specifically, the use of a suitable dimensionality reduction algorithm can reduce these massive data from high dimensions to an acceptable range without losing the meaning expressed by the original data, while the amount of calculation is greatly reduced and easier to understand. In this embodiment, the data cleaning algorithm uses principal component analysis (PCA), which is a commonly used data dimensionality reduction algorithm and is mainly used for high-dimensional data dimensionality reduction processing. Its core idea is to construct new features of m dimensions (m is much smaller than n) and mutually orthogonal through feature extraction on the basis of the original n-dimensional features. The specific process is: first calculate the direction with the largest variance of the original data as the first axis of the new coordinate system, and then continue to look for the direction with the largest variance as the second axis in the direction orthogonal to the axis, and so on to determine the m new feature dimensions. After mapping, most of the variance of the original data is concentrated in the new m-dimensional features, and the variance of the remaining features is almost 0 and can be ignored, thereby achieving dimensionality reduction and simplifying the data structure while retaining the main information of the data. PCA uses orthogonal transformation to convert the observed data represented by linearly correlated variables into a few data represented by linearly independent variables. The linearly independent variables are called principal components; the main calculation process is as follows:

[0066] PCA dimensionality reduction is based on the m-dimensional random variable x i The sample set X is composed of X=(x1,x2,…,x m ) T , we need to find a suitable dimension k (k <= m) to reduce the dimension of the original sample.

[0067]

[0068] Assuming that the average value of each dimension of feature data is 0, E(x) is expressed as the average value of x, and the formula is as follows:

[0069] E(x)=0

[0070] In the process of dimensionality reduction, it is expected that the values ​​of the original data after mapping can achieve the greatest degree of dispersion. Using mathematical knowledge, the variance can be used to represent this degree of dispersion:

[0071]

[0072] Among them, X i is the sample set of the i-th dimension.

[0073] The first task is to determine the first basis vector. Based on the principle that PCA reduces dimensionality by finding the direction with the maximum variance, when all data are projected onto the basis vector, the variance of the projected data should reach the maximum value. This basis vector can capture the difference and change information of the data to the greatest extent. After determining the first basis vector, when looking for the second basis vector, if the direction with the largest variance after projection is still selected, the new direction will most likely coincide with the first direction, and data dimensionality reduction and feature extraction will not be effectively achieved. In order to make the newly constructed feature dimensions independent of each other, so as to more comprehensively extract data features, the second basis vector should be orthogonal to the first basis vector. In this process, the covariance is used to measure the correlation between the two vectors. When the covariance is 0, it indicates that the two vectors are orthogonal to each other, thereby ensuring that the newly selected projection direction meets the requirements of the PCA algorithm.

[0074]

[0075] When the value of Cov(a,b) is equal to 0, it means that the two vectors are completely independent. i ,b i Represents the eigenvector corresponding to the i-th eigenvalue.

[0076] Furthermore, the above data dimensionality reduction process can be summarized as follows:

[0077] 1) Construct a data matrix: Integrate the original data into a matrix X with n rows and m columns.

[0078] 2) Data zero-meaning: First calculate the mean of each row of data in X, and then subtract this mean from each data in the row to achieve data zero-meaning.

[0079] 3) Calculate the covariance matrix: Calculate the covariance matrix based on the zero-mean data In order to reduce the computational complexity of solving the eigenvector in matrix decomposition, the singular value decomposition (Singular Value Decomposition,

[0080] SVD) algorithm.

[0081] 4) Feature solving: Perform feature solving on the covariance matrix C to obtain its eigenvalues ​​and corresponding eigenvectors.

[0082] 5) Screening eigenvectors: Arrange the obtained eigenvalues ​​in descending order, then select the eigenvectors corresponding to the first k larger eigenvalues ​​in turn, and use these vectors to form the matrix P.

[0083] 6) Complete dimensionality reduction: By calculating Y=PX, the data after dimensionality reduction to k dimensions is obtained.

[0084] Through the above process, PCA realizes the mapping of high-dimensional data to k-dimensional space, so that the projected data retains the variance information of the original data as much as possible, thereby effectively reducing the data dimension and improving the efficiency of data processing.

[0085] More specifically, this section uses an example of a state estimation algorithm, specifically the Maximum Correlated Entropy Square Root Cubic Kalman Filter (MCCSCKF) algorithm. The Maximum Correlated Entropy (MCC) criterion is introduced on the basis of SCKF, thereby eliminating the divergence problem in the filter calculation. The following is the basic calculation process of the MCCSCKF algorithm:

[0086] The process model and measurement model are expressed as follows:

[0087]

[0088] Among them, x(k) is the state vector, recorded as x(k)=[v x v y r ψ] T ; u(k) is the control vector, denoted as u(k)=[a x a y δ f ] T ; z(k) is the measurement vector, recorded as z(k)=[rv e,LEO v n,LEO ], w(k) and v(k-1) represent uncorrelated process noise and measurement noise with zero mean, and the covariance matrix is ​​Q(k-1)=E[v(k-1)v T (k-1)],R(k)=E[w(k)w T (k)]. f(.) and h(.) are the nonlinear state and measurement functions of the system.

[0089] The main steps of SCKF are prediction and update, which are as follows:

[0090] 1) Prediction

[0091] Assume that at a certain point k, S(k-1|k-1) is the square root factor of the state error covariance matrix P(k-1|k-1).

[0092] The state error covariance matrix is ​​recorded as:

[0093] P(k-1|k-1)=S(k-1|k-1)S T (k-1|k-1)

[0094] Similarly, S Q (k-1) and S R (k) are the square root factors of the covariance matrix Q(k-1) and R(k). They are respectively denoted as:

[0095]

[0096] Calculate volume point:

[0097]

[0098] Among them, I(i) is the n×n unit matrix, denoted as is the state vector at time k-1. n is the number of volume points.

[0099] Propagation of cube points:

[0100] χ i* (k|k-1)=f(k-1,χ i (k-1|k-1)),i=1,…,2n

[0101] The prior state and square root of the covariance matrix are estimated by,

[0102]

[0103] S(k|k-1)=Tria([X * (k|k-1),S Q (k-1)])

[0104] in, Tria represents the triangular decomposition of a matrix.

[0105] 2) Time update

[0106] Calculate volume point:

[0107]

[0108] Propagation of cube points:

[0109] χ i** (k|k-1)=h(k,χ i (k|k-1)),i=1,…,2n

[0110] The prior measures and square roots of the covariance matrix are estimated by:

[0111]

[0112] in,

[0113] Compute the cross-covariance matrix:

[0114] P xz (k|k-1)=X(k|k-1)Z T(k|k-1)

[0115] in

[0116]

[0117] Calculate the Kalman gain:

[0118]

[0119] The posterior state and square root of the covariance matrix are estimated by:

[0120]

[0121] S(k|k)=Tria([X(k|k-1)-K(k)Z(k|k-1),K(k)S R (k)])

[0122] Due to the excellent performance of correlation entropy in non-Gaussian environments, the MCSCKF algorithm is derived jointly by MCC and SCKF. The specific derivation process of MCSCKF is as follows:

[0123] The nonlinear recursive model is established as:

[0124]

[0125] in, The covariance φ(k) of the matrix can be written as M(k)=M(k)M T (k) Available from obtained by Cholesky decomposition.

[0126] The best estimate of x(k) based on MCC can be obtained as follows,

[0127]

[0128] Where n is the dimension of the state vector, m is the dimension of the measurement vector, and e i (k) is the i-th component of e(k), G σ is the Gaussian kernel function.

[0129] By making the first-order derivative of the above equation zero, we can get the optimal solution of the equation.

[0130]

[0131] Among them, J MCC is the objective function. If we define C i (k) = G σ (ei (k)),

[0132]

[0133] In practice, the true state x(k) is usually unknown. Therefore, the prior measurement noise variance is written as the measurement noise is:

[0134]

[0135] The covariance matrix can be obtained as follows:

[0136]

[0137] In addition, based on the above-mentioned vehicle state estimation system based on a space-ground integrated network, the present invention also proposes a vehicle state estimation method based on a space-ground integrated network, which can complete vehicle state estimation.

[0138] The above embodiments are intended only to illustrate the design concepts and features of the present invention. Their purpose is to enable those skilled in the art to understand the contents of the present invention and implement them accordingly. The scope of protection of the present invention is not limited to the above embodiments. Therefore, any equivalent changes or modifications made based on the principles and design concepts disclosed in the present invention are within the scope of protection of the present invention.

Claims

1. A vehicle state estimation system based on a ground-ground integrated network, characterized in that: The system includes: an on-board LEO signal receiver, a data processor, on-board sensors, a vehicle state estimation system, and a stability control system; The vehicle-mounted LEO signal receiver communicates with a low-orbit satellite and receives satellite data in real time to obtain vehicle information in real time; The data processor cleans the data received by the vehicle-mounted LEO signal receiver; The on-board sensor unit measures the angular velocity and linear acceleration of the vehicle in real time; The vehicle state estimation system estimates the vehicle's center of mass sideslip angle by fusing onboard sensor information and satellite information; The stability control system includes a stability judgment module and an active safety control module. The stability judgment module performs stability judgment based on the center of mass sideslip angle output by the vehicle state estimation system, and the active safety control module activates active safety control based on the stability judgment result.

2. The vehicle state estimation system based on the ground-ground integrated network according to claim 1, characterized in that: In the data processor, the speed in the NE global coordinate is converted to obtain the two-dimensional plane speed v s , recorded as: Among them, v x represents the longitudinal velocity of the vehicle, v y Represents the lateral velocity of the vehicle.

3. The vehicle state estimation system based on the ground-ground integrated network according to claim 1, characterized in that: The vehicle-mounted sensor unit uses a vehicle-mounted IMU to measure the angular velocity and linear acceleration of the vehicle in real time.

4. The vehicle state estimation system based on the ground-ground integrated network according to claim 1, characterized in that: In the vehicle state estimation system, the maximum correlation entropy square root cubic Kalman filter algorithm is used for state estimation.

5. The vehicle state estimation system based on the ground-ground integrated network according to claim 4, characterized in that: The maximum correlation entropy square root cubature Kalman filter method introduces the maximum correlation entropy MCC criterion on the basis of SCKF.

6. The vehicle state estimation system based on the ground-ground integrated network according to claim 5, characterized in that: The specific derivation process of the maximum correlation entropy square root cubature Kalman filter method is as follows: The nonlinear recursive model is established as: Among them, x(k) is the state vector, which is recorded as x(k)=[v x v y r ψ] T ; z(k) is the measurement vector, recorded as z(k)=[rv e,LEO v n,LEO ]; is the state vector at time k-1; h(.) is the measurement function; φ(k) is the process noise nonlinear recursive matrix, denoted as w(k) is the process noise; The best estimate of x(k) based on MCC is obtained as follows, Among them, n is the dimension of the state vector, m is the dimension of the measurement vector, e(k) is the model prediction error, and e i (k) is the i-th component of e(k), G σ is the Gaussian kernel function; By making the first-order derivative of the above equation zero, we can get the optimal solution of the equation. Among them, J MCC is the objective function; if we define C i (k) = G σ (e i (k)), In practice, the true state x(k) is usually unknown. Therefore, the prior measurement noise variance is written as the measurement noise is: The covariance matrix can be obtained as follows: Among them, M r (k) is the weighted square root matrix measuring the prediction error.

7. The vehicle state estimation system based on the ground-ground integrated network according to claim 2, characterized in that: The sideslip angle of the center of mass is as follows:

8. The vehicle state estimation system based on the ground-ground integrated network according to claim 7, characterized in that: In the stability judgment module, the side slip angle β output by the vehicle state estimation system is calculated based on the phase plane. As a basis for stability judgment, when the stability boundary condition of the vehicle yaw-side phase plane is met When , the active safety control unit is not activated, otherwise, the active safety control unit needs to be activated. β is the center of mass side slip angle, is the rate of change of the sideslip angle at the center of mass, k1 and k2 are coefficients related to the road adhesion coefficient and the front wheel angle, and k1 is a dynamic response time constant used to balance the sideslip angle β and its rate of change The weight of reflects the sensitivity of the vehicle system to the rate of change of the sideslip angle; k2 is the boundary amplitude of the stable region, representing the maximum equivalent sideslip energy that the vehicle can withstand.

9. The vehicle state estimation system based on the ground-ground integrated network according to claim 7, characterized in that: The vehicle state estimation system fuses low-orbit satellite data with on-board sensor data to perform state estimation as follows: Among them, I z is the vehicle's moment of inertia, ψ represents the vehicle's yaw angle, r represents the vehicle's yaw angular velocity, and F yf , F yr represents the lateral force of the front and rear tires, l f 、l r Represents the stiffness coefficient of the front and rear tires, are the vehicle's longitudinal velocity v x , the vehicle's lateral velocity v y The first derivative of .

10. A vehicle state estimation method based on a ground-ground integrated network, characterized in that: Vehicle state estimation is performed using the vehicle state estimation system based on a ground-space integrated network as described in claim 1.

Images