Rolling time domain estimation method of vehicle active suspension system

Abstract

A rolling time domain estimation method for a vehicle active suspension system is characterized by comprising the following steps: S1, establishing and initializing a state model of the vehicle active suspension system; s2, a state sensor obtains measurement state information of the vehicle active suspension system; s3, the relay node generates a relay signal from the measurement state information through Kalman filtering, and transmits the relay signal to a rolling time domain estimator; and S4, the rolling time domain estimator calculates state estimation information of the vehicle active suspension system according to the relay signal and the measurement state information. According to the invention, the communication range of wireless communication can be obviously expanded, and the efficient performance of long-distance communication is ensured, so that the accuracy and reliability of the state information of the vehicle active suspension system are improved, and the vehicle active suspension system can be better dynamically adjusted.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle active suspension, and in particular to a rolling time domain estimation method of a vehicle active suspension system. Background Art

[0002] The suspension system is an important system in the car that connects the vehicle to the wheels, absorbs road impact, and maintains vehicle stability and comfort. The vehicle active suspension system is an automotive technology that dynamically and adaptively adjusts the stiffness and damping characteristics of the suspension system according to the driving conditions of the vehicle to achieve the best vibration reduction effect. The state estimation of the vehicle active suspension system refers to the process of using sensor data and mathematical models to determine the current state of the suspension system. These states usually include the dynamic response parameters of the vehicle, such as body position, speed, acceleration, and contact force between the wheels and the road. State estimation is crucial to achieving active control of the suspension system, which allows the system to dynamically adjust the suspension parameters based on real-time data to improve the vehicle's handling, comfort, and safety.

[0003] The existing patent with application number 201810351829.4 discloses an observer-based online estimation method for the roll parameters of a semi-active suspension vehicle. The technical solution is based on a bilinear observer and uses the vertical acceleration measurement values ​​of the sprung mass and the unsprung mass in the multi-degree-of-freedom dynamic model of the lateral and yaw motion of the semi-active suspension vehicle to achieve online observation of the deformation of the left and right tires of the vehicle, thereby estimating the roll state of the vehicle relative to the road surface. The defect of this technical solution is that the real-time state of the active suspension system is difficult to measure online. The estimation method assumes that the state of the active suspension system can be obtained in real time through sensors (bilinear observers), but the installation locations of these sensors are relatively complex and involve key components of the vehicle. In addition, the active suspension system is disturbed by various factors such as vehicle vibration and uneven road surface, resulting in instability of sensor data, making it difficult to obtain the real-time state of the suspension system through online measurement.

[0004] In order to solve the above technical problems, there is a technical solution that uses a rolling time domain estimator to handle the constraints of system state and process interference, such as physical constraints. The rolling time domain estimator is a device for estimating the state of a dynamic system, which has a built-in rolling time domain estimation algorithm and is suitable for vehicle active suspension systems with constraints and time-varying parameters. However, the defect of this technical solution is that since the communication between the sensor and the rolling time domain estimator is wireless communication, the performance of wireless communication is affected and interfered by many factors, such as distance factors, and is unstable, which further leads to signal distortion or signal inaccuracy, making the state information ultimately provided by the rolling time domain estimator unreliable and inaccurate, resulting in incorrect dynamic adjustment of the vehicle active suspension system. Summary of the invention

[0005] In view of the above-mentioned defects of the prior art, the present invention provides a rolling time domain estimation method for a vehicle active suspension system, which can significantly expand the communication range of wireless communication and ensure the high efficiency performance of long-distance communication, thereby significantly improving the communication quality of wireless communication, thereby improving the accuracy and reliability of the status information of the vehicle active suspension system, and further enabling the vehicle active suspension system to be better dynamically adjusted.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] A rolling time domain estimation method for a vehicle active suspension system comprises the following steps:

[0008] S1: Establish and initialize the state model of the vehicle active suspension system;

[0009] S2: The state sensor obtains measurement state information of the vehicle active suspension system;

[0010] S3: The relay node generates a relay signal from the measurement state information through Kalman filtering, and transmits the relay signal to the rolling time domain estimator;

[0011] S4: The rolling time domain estimator calculates state estimation information of the vehicle active suspension system according to the relay signal and the measured state information.

[0012] The state model of the vehicle active suspension system describes the evolution of the system state over time. The state model has multiple model parameters that together determine the behavior of the suspension system. When there is not enough data for an accurate estimate, the initial state can be based on the manufacturer's data or preset conditions. Establish and initialize the dynamic differential equations of the vehicle active suspension system as the state model of the vehicle active suspension system:

[0013]

[0014]

[0015] Among them, M S 、M us denotes sprung mass and unsprung mass respectively; k s is the stiffness coefficient; c s is the damping ratio; z i and q i represent the displacement of the i-th independent suspension in the sprung mass and the unsprung mass respectively; w i and u i They represent the unknown road disturbance and damping force corresponding to the i-th suspension; z represents the vertical displacement of the center of mass of the vehicle body; k t Indicates the stiffness coefficient of the tire; and θ represent the body roll and pitch angles respectively; I x and I y Respectively represent the moment of inertia of roll and pitch; L i Indicates the body torque.

[0016] State sensors are placed at key locations of the vehicle to measure the dynamic response of the vehicle's active suspension system and obtain the measurement state information of the vehicle's active suspension system. The state sensors collect data at a certain sampling frequency. The measurement state information data is real-time, but contains interference from system status and environmental factors, which need to be processed by the rolling time domain estimator. Part of the measurement state information is transmitted through the relay node to generate a relay signal and then received by the rolling time domain estimator, while the other part is directly received by the rolling time domain estimator located in the cloud.

[0017] The relay node receives the measurement state information signal from the state sensor, processes it to generate a relay signal, and then forwards the relay signal to the rolling time domain estimator. The processing method of the relay node is to use the filter forwarding technology. The Kalman filter first removes the influence of factors such as multipath effect, fading, and interference. The Kalman filter is a state estimation algorithm for a linear dynamic system, and its calculation principle and method are existing technologies. The Kalman filter reduces the noise and error in the state sensor data through prediction and update steps, and provides a more accurate relay signal. Selecting the filter forwarding technology of the Kalman filter as the relay forwarding strategy of the relay node can reduce the noise and interference of the signal during the transmission process and enhance the signal quality. At the same time, in wireless communications, the signal is often affected by the distance factor. Setting up relay nodes can effectively expand the coverage of the communication network and improve the signal quality.

[0018] Select filter forwarding technology as the relay forwarding strategy and generate relay signals through Kalman filtering

[0019]

[0020] in is the one-step prediction at time t-1; μ0 is the initial state value; is the filter gain of the Kalman filter algorithm, as follows:

[0021]

[0022] Where P t i and They represent the filtering covariance matrix and the one-step prediction covariance matrix respectively. i and R i As a weight matrix, it is assumed to be positive definite, and Q i and Ri The values ​​are set to and The inverse covariance matrix of .

[0023] Preferably, the step S1 comprises: decomposing the vehicle active suspension system into a plurality of coupled subsystems, wherein the communication network between the subsystems is a directed graph G=(N, E); and the subsystems exchange information through the state sensor N.

[0024] The communication network between subsystems refers to the network structure connecting multiple subsystems, which allows information exchange and data communication between subsystems. In the directed graph G = (N, E), if there are 5 subsystems, then the node N = {1, 2, ..., 5} represents the state sensor, and the element (j, i) in the set E = N × N means that the state sensor j can transmit information to the state sensor i, that is, the subsystems exchange information through the state sensor.

[0025] The state model of the subsystem after decomposition of the vehicle active suspension system is the output equation:

[0026]

[0027] where x i represents the state of subsystem i and i∈N, represents the process disturbance of subsystem i; is the measurement status information of subsystem i, matrix A i , B i and B r i are constant matrices of appropriate dimensions, representing the state vector x of subsystem i i The rate of change over time of itself, the control input and the corresponding process disturbances The correlation of ij ,i=1,2,...,n,j≠i is the subsystem interaction matrix, which represents the dynamic correlation between the jth subsystem and the ith subsystem; represents the measurement noise of subsystem i; C i and D i are constant matrices of appropriate dimensions, representing the correlation between the measured output of the ith subsystem and the state and measurement noise, respectively.

[0028] Preferably, the step S4 comprises: setting a sliding window; and obtaining the state estimation information of the subsystem by minimizing a cost function estimation.

[0029] The calculation method of the rolling horizon estimator is an existing technology, in which the sliding window is used to provide a fixed-length time range to capture the latest state of the vehicle active suspension system in each estimation step. The state of the vehicle active suspension system changes over time and needs to be updated in real time to reflect the latest information. Set a sliding window with a fixed length of N+1 and a range of tN, t-N+1,..., It represents the state of subsystem i at any time t. Estimate of the cost function By minimizing the cost function Estimated value of

[0030]

[0031] in The value of can be calculated by the following formula:

[0032]

[0033] Item 1 It is called the arrival cost, which is assumed to be non-negative. Its role is to serve as prior information, summarizing the impact of the data before time tN on the current data that is not explicitly stated in the cost function; the second term uses the matrix Q i Weighted, reflecting the change of the corresponding subsystem related state over time according to the subsystem state model; the last two items are expressed in matrix R i and The weighted function penalizes the distance between the expected output based on the state estimation and the actual measurement value obtained by the estimator and the measurement value processed by the relay node. In the rolling horizon estimator, the minimization cost function is used to find the estimate that best matches the system state and measurement data within a given sliding window.

[0034] Preferably, the step S4 comprises: adding constraints in the estimation process of minimizing the cost function; the constraints comprise state variable constraints, process disturbance constraints, and measurement noise constraints.

[0035] During the cost function minimization process, some appropriate constraints can be added to reflect physical constraints for certain state variables, as well as process disturbances and measurement noise. The physical constraint of measurement noise takes into account the noise in the measurement process, ensuring that the estimation result will not deviate from the actual state due to abnormal measurement values, ensuring that the solution of the cost function is feasible in actual operation, and can help improve the accuracy of the prediction. State variable constraints limit the possible values ​​of the state variables of the vehicle active suspension system. For example, the displacement of the vehicle suspension cannot exceed the mechanical limit. Process disturbance constraints take into account the external influences that the vehicle active suspension system may be subject to during operation, such as wind resistance and road inequality. Constraints can help stabilize the solution of the optimization problem, especially in the presence of noise and uncertainty.

[0036] The specific constraints are as follows:

[0037] State variable constraints: k=[tN,t]

[0038] Process disturbance constraints: k=[tN,t-1]

[0039] Measurement noise constraints: k=[tN,t]

[0040] Preferably, the step S4 comprises: the subsystem acquires the measurement state information and a priori state estimation information of the neighbor subsystem.

[0041] At time t, due to the coupling relationship between subsystems in the distributed structure, subsystem i can obtain the neighbor subsystem j (j∈N i ) obtains measurements related to the state of the local subsystem i And the prior estimate At this time, the cost function J t i Can be expanded to:

[0042]

[0043] There is a coupling relationship between subsystems. The state of subsystem i will affect the neighboring subsystem j (j∈N i ), and vice versa. Therefore, subsystem i can obtain the measurement data of neighbor subsystem j (j∈N i ) provides these measurement status information, indirectly obtains information about its own state, and thus improves its own state estimation; neighbor subsystem j (j∈N i ) provides indirect information related to the state of subsystem i itself, helping it to make more accurate estimates; prior state estimation information refers to the estimated value obtained by the system based on the previous state and model at time tN;

[0044] The way for a subsystem to obtain the measured state information and prior state estimation information of its neighbor subsystem is to combine the communication topology of the subsystem with a random matrix Combine, the elements inside satisfy:

[0045]

[0046] According to the random matrix and the coupling relationship between subsystems, subsystem i can receive the corresponding information of neighbor subsystems through the communication link.

[0047] Preferably, the step S4 comprises: fusing the local arrival cost of the subsystem, the measurement state information of the neighbor subsystem, and the prior state estimation information of the neighbor subsystem to obtain a local arrival cost.

[0048] Aiming at the distributed structure of vehicle active suspension system and the arrival cost consistency principle in distributed rolling time domain estimation, the local arrival cost and the neighbor subsystem j∈N are combined. i The local arrival cost can be defined by fusing the measurement data of and the prior information of the corresponding state:

[0049]

[0050] The consistency matrix and Satisfy the element requirements of the random matrix and ensure that the fused arrival cost is the weighted average of the local arrival costs; is a positive definite weight matrix used to quantize the prior confidence level.

[0051] Preferably, the step S4 comprises: the subsystem and the neighbor subsystem perform the fusion multiple times at each sampling interval.

[0052] In each sampling interval, multiple (e.g., L) information fusions can be performed between neighbor subsystems. Let l = 1, ..., L, then the consistency fusion algorithm can be expanded to:

[0053]

[0054] and

[0055] When time t>0, the extended consistency fusion algorithm is used to execute L times of the consistency algorithm. Under the constraints of the elements of the random matrix, the state estimation information of the local subsystem i is obtained by minimizing the cost function, and the prior value of the subsystem i at the next moment is obtained by recursion. And the corresponding weight matrix in the target cost function

[0056] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0057] 1. The present invention expands the communication range of wireless communication by introducing relay nodes, and adopts the Kalman filtering method as a relay forwarding strategy, which significantly reduces the impact of noise, thereby improving the accuracy and reliability of the measurement state information. The rolling time domain estimator located in the cloud receives the measurement state information from the state sensor and the relay signal filtered by the relay node and performs state estimation on the suspension system, ensuring the efficient performance of long-distance communication. Ultimately, more accurate and reliable state estimation information can be obtained, so that the vehicle active suspension system can better dynamically adjust parameters to adapt to different driving conditions, thereby improving the vehicle's handling, comfort and safety.

[0058] 2. The distributed rolling time domain estimator that divides the reduced-order subsystem is adopted. By dividing the complex vehicle active suspension system into several reduced-order subsystems, each subsystem only contains a part of the dynamics of the original system, thereby reducing the order and computational complexity of the model. Each subsystem can perform state estimation and data processing independently, reducing the need for global calculations, and can capture local dynamics more finely, improving the accuracy of state estimation. By fusing information between subsystems, the advantages of each subsystem can be combined to improve the overall estimation performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 This is a schematic diagram of the method of Example 1;

[0060] Figure 2 It is a schematic diagram of the lifting posture estimation result of the vehicle active suspension system in Example 1;

[0061] Figure 3 Schematic diagram of pitch attitude estimation of the vehicle active suspension system in Example 1. DETAILED DESCRIPTION

[0062] In order to make the technical means, creative features, objectives and effects of the invention easier to understand, the invention is further described with reference to specific diagrams. However, the invention is not limited to the following implementation cases.

[0063] It should be noted that the structures, proportions, sizes, etc. illustrated in the drawings in this specification are only used to match the contents disclosed in the specification so as to facilitate understanding and reading by persons familiar with this technology. They are not used to limit the conditions under which the present invention can be implemented, and therefore have no substantive technical significance. Any structural modification, change in proportion or adjustment of size, without affecting the effects and purposes that can be achieved by the present invention, should still fall within the scope of the technical contents disclosed by the present invention.

[0064] Embodiment 1:

[0065] like Figure 1 A rolling time domain estimation method for a vehicle active suspension system is shown, comprising the following steps:

[0066] S1: Establish and initialize the state model of the vehicle active suspension system;

[0067] S2: The state sensor obtains the measurement state information of the vehicle active suspension system;

[0068] S3: The relay node generates a relay signal from the measurement state information through Kalman filtering, and transmits the relay signal to the rolling time domain estimator;

[0069] S4: The rolling time domain estimator calculates the state estimation information of the vehicle active suspension system based on the relay signal and the measured state information.

[0070] The state model of the vehicle active suspension system describes the evolution of the system state over time. The state model has multiple model parameters that together determine the behavior of the suspension system. When there is not enough data for an accurate estimate, the initial state can be based on the manufacturer's data or preset conditions. Establish and initialize the dynamic differential equations of the vehicle active suspension system as the state model of the vehicle active suspension system:

[0071]

[0072]

[0073] Among them, M S 、M us denotes sprung mass and unsprung mass respectively; k s is the stiffness coefficient; c s is the damping ratio; z i and q i represent the displacement of the i-th independent suspension in the sprung mass and the unsprung mass respectively; w i and u i They represent the unknown road disturbance and damping force corresponding to the i-th suspension; z represents the vertical displacement of the center of mass of the vehicle body; k t Indicates the stiffness coefficient of the tire; and θ represent the body roll and pitch angles respectively; I x and I y Respectively represent the moment of inertia of roll and pitch; L i Indicates the body torque;

[0074] State sensors are placed at key locations on the vehicle to measure the dynamic response of the vehicle's active suspension system and obtain the measured state information of the vehicle's active suspension system. The state sensors collect data at a certain sampling frequency. The measured state information data is real-time, but contains interference from system states and environmental factors, which need to be processed by a rolling time domain estimator.

[0075] The relay node receives the measurement state information signal from the state sensor, processes it to generate a relay signal, and then forwards the relay signal to the rolling time domain estimator. The processing method of the relay node is to use the filter forwarding technology. The Kalman filter first removes the influence of factors such as multipath effect, fading, and interference. The Kalman filter is a state estimation algorithm for a linear dynamic system, and its calculation principle and method are existing technologies. The Kalman filter reduces the noise and error in the state sensor data through prediction and update steps, and provides a more accurate relay signal. Selecting the filter forwarding technology of the Kalman filter as the relay forwarding strategy of the relay node can reduce the noise and interference of the signal during the transmission process and enhance the signal quality. At the same time, in wireless communications, the signal is often affected by the distance factor. Setting up relay nodes can effectively expand the coverage of the communication network and improve the signal quality.

[0076] Select filter forwarding technology as the relay forwarding strategy and generate relay signals through Kalman filtering

[0077]

[0078] in is the one-step prediction at time t-1; μ0 is the initial state value; is the filter gain of the Kalman filter algorithm, as follows:

[0079]

[0080] Where P t i and They represent the filtering covariance matrix and the one-step prediction covariance matrix respectively. i and R i As a weight matrix, it is assumed to be positive definite, and Q i and R i The values ​​are set to and The inverse covariance matrix of .

[0081] In summary, the present invention expands the communication range of wireless communication by introducing relay nodes, and adopts the Kalman filtering method as a relay forwarding strategy, which significantly reduces the impact of noise, thereby improving the accuracy and reliability of the measurement state information. The rolling time domain estimator located in the cloud receives the measurement state information from the state sensor and the relay signal filtered by the relay node and performs state estimation on the suspension system, ensuring the efficient performance of long-distance communication. Ultimately, more accurate and reliable state estimation information can be obtained, so that the vehicle active suspension system can better dynamically adjust parameters to adapt to different driving conditions, thereby improving the vehicle's handling, comfort and safety.

[0082] Step S1 also includes: decomposing the vehicle active suspension system into multiple coupled subsystems, wherein the communication network between the subsystems is a directed graph G=(N, E); and the subsystems exchange information through a state sensor N.

[0083] The communication network between subsystems refers to the network structure connecting multiple subsystems, which allows information exchange and data communication between subsystems. In the directed graph G = (N, E), if there are 5 subsystems, then the node N = {1, 2, ..., 5} represents the state sensor, and the element (j, i) in the set E = N × N means that the state sensor j can transmit information to the state sensor i, that is, the subsystems exchange information through the state sensor.

[0084] The state model of the subsystem after decomposition of the vehicle active suspension system is the output equation:

[0085]

[0086] where x i represents the state of subsystem i and i∈N, represents the process disturbance of subsystem i; is the measurement status information of subsystem i, matrix A i , B i and are constant matrices of appropriate dimensions, representing the state vector x of subsystem i i The rate of change over time of itself, the control input and the corresponding process disturbances The correlation of ij ,i=1,2,...,n,j≠i is the subsystem interaction matrix, which represents the dynamic correlation between the jth subsystem and the ith subsystem; represents the measurement noise of subsystem i; C i and D i are constant matrices of appropriate dimensions, representing the correlation between the measured output of the ith subsystem and the state and measurement noise, respectively.

[0087] In the vehicle active suspension system of this embodiment, there is a strong coupling relationship between the four independent suspensions of the vehicle and the body posture. The body posture (such as pitch and lift) is closely related to the dynamic behavior of each suspension. The force and movement of the suspension will affect the overall posture of the body, and the change of the body posture will in turn affect the working state of each suspension. In order to reduce the complexity, according to the mutual influence between the suspension and the body posture, the system is decomposed into five coupled reduced-order subsystems, each suspension is a subsystem, the vehicle center is the fifth subsystem, and the subsystems are coupled to each other.

[0088] Specifically, step S4 further includes: setting a sliding window; and obtaining state estimation information of the subsystem by minimizing the cost function estimation of the subsystem.

[0089] The calculation method of the rolling horizon estimator is an existing technology, in which the sliding window is used to provide a fixed-length time range to capture the latest state of the vehicle active suspension system in each estimation step. The state of the vehicle active suspension system changes over time and needs to be updated in real time to reflect the latest information. Set a sliding window with a fixed length of N+1 and a range of tN, t-N+1,..., It represents the state of subsystem i at any time t. Estimate of the cost function By minimizing the cost function Estimated value of

[0090]

[0091] in The value of can be calculated by the following formula:

[0092]

[0093] Item 1 It is called the arrival cost, which is assumed to be non-negative. Its role is to serve as prior information, summarizing the impact of the data before time tN on the current data that is not explicitly stated in the cost function; the second term uses the matrix Q i Weighted, reflecting the change of the corresponding subsystem related state over time according to the subsystem state model; the last two items are expressed in matrix R i and The weighted function penalizes the distance between the expected output based on the state estimation and the actual measurement value obtained by the estimator and the measurement value processed by the relay node. In the rolling horizon estimator, the minimization cost function is used to find the estimate that best matches the system state and measurement data within a given sliding window.

[0094] In step S4, constraints are added in the estimation process of minimizing the cost function; the constraints include state variable constraints, process disturbance constraints, and measurement noise constraints.

[0095] During the cost function minimization process, some appropriate constraints can be added to reflect physical constraints for certain state variables, as well as process disturbances and measurement noise. The physical constraint of measurement noise takes into account the noise in the measurement process, ensuring that the estimation result will not deviate from the actual state due to abnormal measurement values, ensuring that the solution of the cost function is feasible in actual operation, and can help improve the accuracy of the prediction. State variable constraints limit the possible values ​​of the state variables of the vehicle active suspension system. For example, the displacement of the vehicle suspension cannot exceed the mechanical limit. Process disturbance constraints take into account the external influences that the vehicle active suspension system may be subject to during operation, such as wind resistance and road inequality. Constraints can help stabilize the solution of the optimization problem, especially in the presence of noise and uncertainty.

[0096] The specific constraints are as follows:

[0097] State variable constraints:

[0098] Process disturbance constraints:

[0099] Measurement noise constraints:

[0100] In step S4, the subsystem obtains the measurement state information and prior state estimation information of the neighbor subsystem.

[0101] At time t, due to the coupling relationship between subsystems in the distributed structure, subsystem i can obtain the neighbor subsystem j (j∈N i ) obtains measurements related to the state of the local subsystem i And the prior estimate At this time, the cost function Can be expanded to:

[0102]

[0103] There is a coupling relationship between subsystems. The state of subsystem i will affect the neighboring subsystem j (j∈N i ), and vice versa. Therefore, subsystem i can obtain the measurement data of neighbor subsystem j (j∈N i ) provides these measurement status information, indirectly obtains information about its own state, and thus improves its own state estimation; neighbor subsystem j (j∈N i) provides indirect information related to the state of subsystem i itself, helping it to make more accurate estimates; prior state estimation information refers to the estimated value obtained by the system based on the previous state and model at time tN;

[0104] The way for a subsystem to obtain the measured state information and prior state estimation information of its neighbor subsystem is to combine the communication topology of the subsystem with a random matrix Combine, the elements inside satisfy:

[0105]

[0106]

[0107]

[0108] According to the random matrix and the coupling relationship between subsystems, subsystem i can receive the corresponding information of neighbor subsystems through the communication link.

[0109] In step S4, the local arrival cost of the subsystem, the measured state information of the neighbor subsystem, and the prior state estimation information of the neighbor subsystem are integrated to obtain the local arrival cost.

[0110] Aiming at the distributed structure of vehicle active suspension system and the arrival cost consistency principle in distributed rolling time domain estimation, the local arrival cost and the neighbor subsystem j∈N are combined. i The local arrival cost can be defined by fusing the measurement data of and the prior information of the corresponding state:

[0111]

[0112] The consistency matrix and Satisfy the element requirements of the random matrix and ensure that the fused arrival cost is the weighted average of the local arrival costs; is a positive definite weight matrix used to quantize the prior confidence level.

[0113] Furthermore, the subsystem is fused with its neighbor subsystems multiple times at each sampling interval.

[0114] In each sampling interval, multiple (e.g., L) information fusions can be performed between neighbor subsystems. Let l = 1, ..., L, then the consistency fusion algorithm can be expanded to:

[0115]

[0116] and

[0117] When time t>0, the extended consistency fusion algorithm is used to execute L times of the consistency algorithm. Under the constraints of the elements of the random matrix, the state estimation information of the local subsystem i is obtained by minimizing the cost function, and the prior value of the subsystem i at the next moment is obtained by recursion. And the corresponding weight matrix in the target cost function

[0118] In this embodiment, the estimation process of the vehicle active suspension system is as follows:

[0119] 1. The sampling period of the vehicle status sensor is T s =0.04s, the initial state is: q1(t0)=-0.02m, q2(t0)=-0.02m, q3(t0)=0.02m, q4(t0)=0.02m, the initial displacement of the suspension center in the vertical direction z(t0)=0m, the initial pitch angle is θ(t0)=0.02°, the initial roll angle is φ(t0)=0.02°, the state sensor collects the dynamic data of the vehicle suspension, including the vertical vibration displacement, pitch angle, roll angle, etc. of the vehicle body, and transmits the data to the relay node through the vehicle communication system or directly to the estimator of the cloud computing center.

[0120] 2. The relay node is located around the vehicle or other suitable locations. After receiving the vehicle data, the relay node obtains the relay signal after filtering according to the Kalman filter. The filtered signal is then forwarded to the estimator in the cloud computing center through wireless communication technology for state estimation.

[0121] 3. After receiving the data from the vehicle-mounted sensors and the signals filtered by the relay nodes, the estimator in the cloud computing center uses the rolling time domain estimation method to process the state and noise constraints of the system; the extended consistency fusion algorithm is used to execute the consistency algorithm L times, and the estimated value of subsystem i is obtained by minimizing the cost function. Recursively obtain the prior value of subsystem i at the next moment And the corresponding weight matrix in the target cost function The above process is repeated to realize real-time state estimation of the lifting, pitching and rolling posture of the vehicle suspension system. Figure 2 This is the result diagram of the vehicle active suspension system lifting posture estimation. Figure 3 This is the pitch attitude estimation result diagram of the vehicle active suspension system.

Claims

1. A rolling time domain estimation method for a vehicle active suspension system, characterized in that: The following steps are involved: S1: Establish and initialize the state model of the vehicle active suspension system; S2: The state sensor obtains measurement state information of the vehicle active suspension system; S3: The relay node generates a relay signal from the measurement state information through Kalman filtering, and transmits the relay signal to the rolling time domain estimator; S4: The rolling time domain estimator calculates state estimation information of the vehicle active suspension system according to the relay signal and the measured state information.

2. The rolling time domain estimation method of the vehicle active suspension system according to claim 1, characterized in that: The step S1 comprises: decomposing the vehicle active suspension system into a plurality of coupled subsystems, wherein the communication network between the subsystems is a directed graph G=(N, E); and the subsystems exchange information through the state sensors.

3. The rolling time domain estimation method of the vehicle active suspension system according to claim 2, characterized in that: The step S4 comprises: setting a sliding window; and obtaining the state estimation information of the subsystem by minimizing a cost function estimation of the subsystem.

4. The rolling time domain estimation method of the vehicle active suspension system according to claim 1, characterized in that: The step S4 includes: adding constraints in the estimation process of minimizing the cost function; the constraints include state variable constraints, process interference constraints, and measurement noise constraints.

5. The rolling time domain estimation method of the vehicle active suspension system according to claim 3, characterized in that: The step S4 includes: the subsystem acquires the measurement state information and prior state estimation information of the neighbor subsystem.

6. The rolling time domain estimation method of the vehicle active suspension system according to claim 5, characterized in that: The step S4 includes: fusing the local arrival cost of the subsystem, the measurement state information of the neighbor subsystem, and the prior state estimation information of the neighbor subsystem to obtain a local arrival cost.

7. The rolling time domain estimation method of the vehicle active suspension system according to claim 6, characterized in that: The step S4 includes: the subsystem and the neighbor subsystem perform the fusion multiple times at each sampling interval.

Images