Over-the-horizon active suspension preview control method based on high-precision positioning pavement state recognition

Abstract

The invention discloses a beyond-visual-range active suspension preview control method based on high-precision positioning road surface state recognition, and the method comprises the steps: carrying out the positioning of a vehicle based on a combined positioning system of low-orbit satellite positioning, a high-precision map and an inertial navigation system, and obtaining the position information of the vehicle; road surface state recognition is carried out based on the position information of the vehicle, and the road surface state comprises a road surface grade and pulse excitation; signal interaction between the vehicles is established, and interactive transmission of road surface states, suspension parameters and vehicle positioning information between the vehicle and surrounding vehicles is achieved; formulating a beyond-visual-range preview control strategy according to the received road surface state, suspension parameters and vehicle positioning information; if the pulse excitation does not exist in the road section, the vehicle suspension controller adopts fuzzy control; if the pulse excitation exists, the vehicle suspension controller adopts LQG control containing preview information; the method can effectively solve the problems that in the preview control process, the sensing range is limited, and the calculation capacity of a vehicle-mounted electronic control unit is insufficient.

Description

Technical Field

[0001] The invention relates to the field of vehicle dynamics and control technology, and in particular to a beyond-visual-range active suspension preview control method based on high-precision positioning road surface state recognition. Background Art

[0002] With the rapid development of the automobile industry, people's requirements for vehicle driving comfort and handling stability are increasing day by day. Especially for freight vehicles, ensuring the integrity of goods is the main purpose during transportation. Existing freight vehicles are mostly based on traditional passive suspension, but due to its inherent characteristics, it is difficult to meet the smoothness requirements under different road conditions at the same time. Therefore, active suspension technology came into being. It can adjust the suspension parameters in real time according to the vehicle's driving status to improve the driving experience.

[0003] However, active suspension based on feedback control will have a certain lag. When facing complex and changeable road conditions, especially some sudden potholes, bumps or continuous bumpy roads, it is impossible to make accurate and effective adjustments in advance, which greatly reduces the driving comfort of the vehicle and even affects the control safety. At this time, the vehicle needs to be able to know the road information ahead in advance to ensure the real-time effect of suspension control. However, most of the existing active suspension preview controls are limited by the performance of the sensor technology, and it is difficult to accurately obtain the road attribute information ahead of the vehicle, resulting in limited perception of the vehicle. On the other hand, with the increase in the amount of data collected by sensors such as cameras and lidars and the complexity of advanced application control strategies, a large amount of computing and storage resources are required, and the storage and computing capabilities of the on-board electronic control unit are limited, which can easily cause bottlenecks in road perception and data processing, making it difficult to meet the needs of high-quality control of the active suspension system.

[0004] Therefore, in order to solve the problems of limited perception range and insufficient computing power of the on-board electronic control unit during preview control, it is urgent to achieve beyond-visual-range road condition perception and efficient preview control of the active suspension. Summary of the invention

[0005] In order to solve the deficiencies in the prior art, the present application proposes a beyond-visual-range active suspension preview control method based on high-precision positioning road state recognition, which can effectively solve the problems of limited perception range and insufficient computing power of the on-board electronic control unit during the preview control process.

[0006] The technical solution adopted by the present invention is as follows:

[0007] The beyond-visual-range active suspension preview control method based on high-precision positioning road state recognition includes the following steps:

[0008] Step 1: Based on the combined positioning system of low-orbit satellite positioning, high-precision map and inertial navigation system, the vehicle is positioned to obtain the vehicle's location information;

[0009] Step 2: Based on the vehicle's location information, the road surface state is identified, and the road surface state includes the road surface grade and pulse excitation; and signal interaction between vehicles is established to achieve interactive transmission of road surface state, suspension parameters, and vehicle positioning information between the vehicle and surrounding vehicles;

[0010] Step 3: Based on the received road surface conditions, suspension parameters, and vehicle positioning information, a beyond-visual-range preview control strategy is formulated; if there is no pulse excitation on the road section, the vehicle suspension controller adopts fuzzy control; if there is a pulse excitation, the vehicle suspension controller adopts LQG control containing preview information.

[0011] Furthermore, during the process of positioning the vehicle, the combined positioning system uses the inertial navigation system to obtain the vehicle mileage, the high-precision map provides candidate roads based on the vehicle mileage, and the low-orbit satellite provides the distance from the vehicle to the candidate road and the angle between the vehicle heading and the direction of the candidate road. The real positioning information of the vehicle is obtained in the map through fusion calculation.

[0012] Furthermore, the method for determining the real positioning information of the vehicle by fusion calculation is as follows: after obtaining the distance d from the vehicle to the candidate road i and the angle ξ between the vehicle heading and the candidate road direction i Finally, the distribution probability corresponding to each candidate road is calculated using Gaussian distribution; the candidate road corresponding to the maximum distribution probability is selected, and the vehicle road and position are located using the weighted matching value.

[0013] Further, the road surface grade identification method is:

[0014] ① Obtaining suspension response signals, including: collecting the vehicle's vertical acceleration signal through an acceleration sensor, solving the suspension dynamic deflection based on the vehicle dynamics equation, and obtaining the vehicle's body speed through a low-orbit satellite;

[0015] ② Preprocessing of suspension response signals: The vehicle speed is divided into grades, and the road surface grade is graded according to the road conditions, so as to traverse the conditions of the vehicle driving on different road surfaces at each speed gear and obtain the vehicle response;

[0016] ③Signal decomposition: Perform variational modal decomposition on the suspension dynamic deflection and sprung mass acceleration signals in the vehicle response signal, and extract characteristic components respectively; and select characteristic components that can obviously contain road surface grade information, so that the characteristic components can obviously contain road surface information;

[0017] ④ Signal feature calculation: reconstruct the selected characteristic components containing road surface information to form an initial feature matrix, and calculate the singular value entropy of the matrix;

[0018] ⑤ Road surface grade classification: Rely on machine learning to automatically establish fuzzy rules, obtain the corresponding relationship between singular value entropy and road surface grade, and realize road surface grade identification based on the response signal.

[0019] Further, confirm the method of pulse excitation:

[0020] The location where the pulse excitation is generated on the road surface is confirmed through a high-precision map. The vehicle in front that passes the location uploads the location information to the cloud to correct the map. At the same time, the location information is synchronously sent to the following vehicle. The vehicle ECU of the following vehicle calculates the time it takes for the vehicle to arrive at the pulse excitation location based on the vehicle speed information provided by the low-orbit satellite and its own positioning information.

[0021] Furthermore, the method for establishing signal interaction between vehicles is:

[0022] Based on the positioning information provided by the combined positioning system, after determining the position of the paired vehicles and the distance between the two vehicles, the base station is found in the area between the two vehicles and the base station is used to transmit information.

[0023] Further, the fuzzy control process is as follows:

[0024] 1. The input of the controller is defined as the difference e and the change rate ec between the suspension dynamic deflection and the ideal value in the suspension system;

[0025] 2. Select the basic domain and quantization factor of the input signal and output signal respectively. The output signal can select the output upper limit of the suspension actuator as its domain range. The input signal domain selects the response range of the suspension dynamic deflection on a C-grade road surface at a speed of 50km / h, and uses this as a standard to calculate the quantization factor value K under different road surfaces at different speeds. i ;

[0026] 3. Design membership function: The input variables are e and ec, and 7 linguistic fuzzy subsets {NB (negative large), NM (negative medium), NS (negative small), Z (zero), PS (positive small), PM (positive medium), PB (positive large)} are used. The membership functions are all trimf-shaped;

[0027] 4. Design fuzzy rules:

[0028] (4) When the values ​​of e and ec are large and in the same direction, the system error is large and the error is still increasing.

[0029] The system needs to provide a large control amount to reduce its error and ensure the stability of the system;

[0030] (5) When the values ​​of e and ec are large and in opposite directions, it means that the system error is large but decreasing. The system needs to try to ensure that the control amount is not too large, that the system does not produce large oscillations, and that the output of the control force is stable.

[0031] (6) When the values ​​of e and ec are small, the system begins to approach stability. When the input value is in the same direction, a very small output force should be selected. When the input value is in the opposite direction, the output force can be kept at zero to prevent the system from oscillating.

[0032] Further, the LQG control method is as follows:

[0033] Construct the vehicle dynamics equation and rewrite it into a state equation, recorded as:

[0034]

[0035] y=Cx+Du

[0036] Let x1 = Z uf 、x2=Z sf 、x3=Z ur 、x4=Z sr , so

[0037] The state variables are:

[0038] The output variables are:

[0039] The control input is:

[0040] The noise input is: w = [x 01 x 02 ];

[0041] Among them, A, B, C, D, E are coefficient matrices, x 01 is the road surface input passed by the front wheel, x 02 is the road surface input of the rear wheels; Z uf , Z ur is the vertical displacement of the front and rear tires of the vehicle; Z sf , Z sr F is the vertical displacement of the front and rear body of the vehicle; f 、F r Output force for its front and rear active suspension actuators;

[0042] The vehicle body acceleration and front suspension dynamic deflection are used to evaluate the driver's riding comfort, the rear suspension vertical acceleration and rear suspension dynamic deflection are used to evaluate the cargo integrity, and the front and rear tire dynamic loads are used to evaluate the road friendliness. Considering the limitation of control energy, the performance index functional of the active suspension optimal controller is defined as:

[0043]

[0044] Among them, q1, q2, q3, q4, q5, q6, r1 and r2 are weighted coefficients of each indicator; is the vertical acceleration of the vehicle's center of mass, and t is the time variable of the time domain signal.

[0045] Introducing the output variable, the indicator functional is further transformed into:

[0046]

[0047] In the formula, X is the state variable of the system, Q is the weight coefficient matrix of the state variable, N is the cross-term weight coefficient matrix; R is the control variable weight coefficient matrix; U is the control vector of the system.

[0048] The optimal control vector of the system is obtained as:

[0049] U=-KX=-R -1 (N T +B T L)X

[0050] Where K is the feedback gain matrix of LQG control, B is the coefficient matrix of system control, and L is the solution of the Riccati equation.

[0051] After combining the additional state vector with the system state space equation, the system state equation combined with preview control and with the additional state vector is:

[0052]

[0053] In the formula, is the first-order derivative of the system’s state variables, is the first-order derivative of the additional state vector, A is the coefficient matrix of the system, F is the coefficient matrix of the additional state vector, A η , D η 、E η , B η are the coefficient matrices of the system after adding the state vector, respectively a0 and a1 are coefficients, respectively denoted as coefficients coefficient

[0054] The above system state equation is combined with the previous performance index functional equation J to solve the Riccati equation, and combined with the optimal control gain matrix of the wheelbase preview information, the optimal control force of the wheelbase preview can be obtained:

[0055] U p (t) = K p [X(t)+η(t)]

[0056] Among them, Kp is the optimal control gain matrix combined with the wheelbase preview information, X(t) is the system state variable in the time domain, and η(t) is the additional state vector in the time domain.

[0057] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the above-mentioned beyond-visual-range active suspension preview control method based on high-precision positioning road state recognition when executing the computer program.

[0058] A computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-mentioned beyond-visual-range active suspension preview control method based on high-precision positioning road surface state recognition.

[0059] Beneficial effects of the present invention:

[0060] (1) The present invention utilizes low-orbit satellite positioning, high-precision maps and inertial navigation system fusion technology to accurately locate the vehicle in real time, so that the positioning accuracy reaches the centimeter level or even higher. Not only can the detailed road information in front of the vehicle be obtained, but the position of the front and rear vehicles can also be accurately located. At the same time, the precise position of the vehicle is fed back to the roadbed base station in real time, which facilitates the roadbed base station to carry out targeted signal transmission based on the vehicle distribution, optimize the configuration of communication resources, and improve communication efficiency. In addition, combined with the vehicle's driving trajectory and time series information, the pulse excitation position information generated by speed bumps, potholes and other factors during the driving of the front and rear vehicles is analyzed, providing key data support for subsequent road state identification and suspension control strategy formulation.

[0061] (2) The present invention establishes signal interaction between vehicles. On the one hand, the vehicle sends information such as the road surface grade, suspension parameters, and self-positioning to the vehicle behind it in real time, and receives similar information from adjacent vehicles at the same time, so as to have a more comprehensive understanding of the road surface feedback status under local road conditions; on the other hand, it maintains close communication with road-side facilities (such as intelligent transportation base stations, roadside units, etc.), receives more detailed road surface monitoring data and traffic flow information from road-side sensors to correct its own status, and uploads data on the road surface it has passed through for the road-side to upgrade the map of the area. With the help of road-side facilities, the above data is transmitted and shared in real time among all vehicles in the area, and the actual working status of the suspension is fed back to the control center, so as to optimize and adjust the control strategy in real time, form a closed-loop control, ensure the stability and reliability of the system, and complete the intelligent monitoring and control management of vehicles and roads on the entire road section.

[0062] (3) The present invention obtains feedback information about the road surface that the preceding vehicle passes through by communicating with the preceding vehicle, including detailed data such as the road surface grade and pulse excitation position. On the one hand, this information is transmitted to the suspension system of the following vehicle in real time, so that the following vehicle can know the road surface conditions ahead in advance, so as to adjust the suspension parameters in time and improve driving comfort; on the other hand, the high-precision map is dynamically updated and enriched using these real-time collected data to make up for the problem of information lag or missing in the map. Especially in special environments such as tunnels, the accuracy of high-precision map information may be reduced due to limited satellite signals. At this time, the road surface information of the preceding vehicle collected by the road surface state recognition module is particularly critical, and the road surface information in the tunnel can be accurately corrected, such as identifying new potholes and waterlogged areas in the tunnel, to ensure safe and smooth driving of the vehicle in complex and special environments. For road surface state related information obtained from different sources, an adaptive weighted fusion algorithm is used to dynamically allocate weights according to the real-time accuracy, reliability and relevance of the information to the current vehicle driving scene, so as to improve the accuracy of road surface state recognition.

[0063] (4) The present invention utilizes the fusion of vehicle sensors and high-precision maps to achieve beyond-visual-range recognition of the road conditions ahead while reducing vehicle costs, providing sufficient road information for active suspension control in advance, overcoming the problems of limited perception range and insufficient computing power of the on-board electronic control unit during preview control, significantly improving the adaptability of the vehicle under various road conditions, and providing reliable data for the control of subsequent vehicles passing through the road section.

[0064] (5) The present invention takes into account the integrity of the cargo, adjusts the control strategy according to the actual working conditions of the suspension and the road conditions ahead, and takes into account random road surfaces and pulse excitations. The impact of the two excitations on the vehicle chassis is very different. Separate processing can better utilize the information obtained by the preview system, making the preview control effect better. The control effect and real-time performance of the active suspension system are ensured, so that the entire system is always in the best working state, extending the service life of the system, reducing maintenance costs, and ensuring cargo transportation efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 It is a combined positioning flow chart.

[0066] Figure 2 It is a vehicle-base station data interaction system.

[0067] Figure 3 This is a schematic diagram of the 1 / 2 car dynamics model.

[0068] Figure 4 It is a flow chart of the road surface grade identification method.

[0069] Figure 5 This is a schematic diagram of the over-the-wheelbase preview signal transmission. Figure 6 It is a schematic diagram of the beyond-visual-range preview control strategy. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is 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 used to limit the present invention.

[0071] Combined with Figure 1-5 The beyond-visual-range active suspension preview control method based on high-precision positioning road state recognition proposed by the present invention has the following specific steps:

[0072] Step 1: The vehicle is positioned through a combined positioning system of low-orbit satellite positioning, high-precision maps and inertial navigation systems, which can be used to obtain the specific location of the vehicle in front, behind or surrounding vehicles. Specifically, in the combined positioning system, the inertial navigation system provides the vehicle mileage, the high-precision map provides candidate roads based on the vehicle mileage calculation results, and the low-orbit satellite provides coordinate points and vehicle heading, speed and other information; through the fusion calculation of the three, the real positioning information of the vehicle can be obtained in the map; and the obtained real positioning information can be used to correct the output of the inertial navigation system to eliminate errors. The combined positioning process is as follows: Figure 1 As shown, the details are as follows:

[0073] (1) The specific process of determining vehicle mileage by the inertial navigation system is as follows:

[0074] After the vehicle starts, the dead reckoning (DR) uses the attitude, heading and mileage information to calculate the relative position of the vehicle relative to the starting point. The signal output by the odometer is generally the incremental distance traveled by the vehicle in a short period of time. The odometer output can be regarded as the instantaneous speed. The DR of the data obtained using the odometer can be described in the following way:

[0075]

[0076] Here, x0 and y0 are set as the initial coordinates of the x-axis and y-axis respectively. At this time, θ0 is the initial heading angle output by the inertial navigation system installed on the vehicle, and x k is the coordinate position of the vehicle in the x-axis direction after k segments of displacement, y k is the coordinate position of the vehicle in the y-axis direction after passing through k segments of displacement; Δθ j is the vehicle heading angle difference between two adjacent displacement segments, θ k is the vehicle heading angle after k-stage displacement, where Δθ k-1 S is the difference in the vehicle heading angle from the k-1 displacement to the k displacement; iis the displacement of the vehicle in the i-th time period (a period of time). i It is calculated from the pulse signal obtained from the non-steering wheel of the vehicle:

[0077]

[0078] Among them, ΔN i is the output increment of the odometer in the i-th sampling period, R is the resolution of the odometer, and D is the diameter of the wheel.

[0079] However, the error of dead reckoning will gradually increase with the accumulation of distance. In order to reduce the maximum error and average error of dead reckoning and improve the vehicle positioning accuracy, the present invention introduces high-precision maps and low-orbit satellites to correct the vehicle's position information.

[0080] (2) Relying on high-precision maps to provide multiple candidate roads for the vehicle's location, the candidate road range of the vehicle's area must be determined. While ensuring that the correct path is included, the number and length of the roads to be matched must be minimized to ensure matching accuracy and efficiency. According to the vehicle position calculated by dead reckoning, all paths within the surrounding area are divided. Combined with the heading angle output by dead reckoning, roads with large errors are eliminated to finally obtain the remaining multiple candidate roads.

[0081] (3) Using low-orbit satellites to provide the distance d from the vehicle to the candidate road i and the angle ξ between the vehicle heading and the candidate road direction i The specific process is as follows:

[0082] The coordinates M of two consecutive points on the road are provided by a low-orbit satellite i1 (x i1 ,y i1 ) and N i1 (x i2 ,y i2 ), the vehicle’s position at this time is set to P(x0,y0), and the distance from the vehicle to the candidate road can be expressed as:

[0083]

[0084] Vehicle heading β P With candidate road directions The angle can be expressed as:

[0085]

[0086] (4) The information obtained from low-orbit satellite positioning, high-precision maps and inertial navigation systems is integrated and calculated to obtain the vehicle's real positioning information in the map; the obtained positioning information can also be used to correct the output of the inertial navigation system itself.

[0087] The method for determining the real positioning information of the vehicle by fusion calculation is as follows: after obtaining the distance d from the vehicle to the candidate road i and the angle ξ between the vehicle heading and the candidate road direction i Then, the Gaussian distribution is used to calculate the distribution probability corresponding to each candidate road. and Since the larger the value of the distribution probability p is, the higher the matching degree is, the candidate road corresponding to the maximum distribution probability is selected and recorded as the candidate road with the highest matching degree.

[0088] Based on the selected candidate roads, the weighted matching value is finally used Locate the vehicle road and position. Among them, α1 and α2 are the distance and angle weight coefficients respectively.

[0089] Step 2: Based on the vehicle's position information, the road surface condition is identified, and the road surface condition includes road surface grade and pulse excitation; and signal interaction is established between vehicles to realize the interactive transmission of road surface condition, suspension parameters, and vehicle positioning information between the vehicle and surrounding vehicles.

[0090] In this embodiment, in order to reduce the cost of road condition identification, the present invention integrates high-precision maps, low-orbit satellites and vehicle dynamic response inverse measurement methods to perform road condition identification. The road condition mainly includes road grade and pulse excitation.

[0091] In this embodiment, combined with Figure 4 As shown, the specific process of the road surface grade identification method of the present invention is as follows:

[0092] ① Obtain suspension response signal: Collect the vehicle's vertical acceleration signal through the acceleration sensor, solve the suspension dynamic deflection based on the vehicle dynamics equation, and rely on low-orbit satellites to obtain signals such as the vehicle's body speed.

[0093] In this embodiment, a four-wheeled pickup truck is used as an example to establish a 1 / 2 vehicle model. Figure 3 As shown, the vehicle dynamics equation is constructed. The vehicle dynamics equation includes the differential equation of vertical motion dynamics at the center of mass of the vehicle body, the differential equation of vehicle body pitch motion dynamics, and the differential equation of vertical motion dynamics of the unsprung mass (front and rear), which are as follows:

[0094] The differential equation of vertical motion dynamics at the center of mass of the vehicle body:

[0095]

[0096] The differential equation of vehicle body pitch motion dynamics:

[0097]

[0098] Differential equation of front wheel vertical motion dynamics:

[0099]

[0100] The differential equation of the vertical motion dynamics of the rear wheel is:

[0101]

[0102] Among them, M s is the sprung mass of the vehicle, M uf is the mass of the front wheel of the vehicle; M ur is the mass of the rear wheel of the vehicle; θ is the pitch angle of the vehicle body; is the vehicle body pitch angular acceleration; I is the vehicle body moment of inertia; a and b are the distances from the front and rear axles of the vehicle to the center of mass of the vehicle; K tf , K tr is the stiffness of the front and rear tires, ignoring damping, K f , K r is the stiffness of the front and rear suspension; C f , C r is the damping of the front and rear suspension; F f 、F r is the output force of the front and rear active suspension actuators; Z s is the vertical displacement of the vehicle center of mass, is the vertical acceleration of the vehicle's center of mass; Z sf , Z sr is the vertical displacement of the front and rear body of the vehicle; Z is the vertical velocity of the vehicle front and rear; uf , Z ur is the vertical displacement of the front and rear tires of the vehicle; is the vertical speed of the front and rear tires of the vehicle; is the vertical acceleration of the front and rear tires of the vehicle; Z f , Z r Motivate for both front and rear roads.

[0103] ② Preprocessing of suspension response signal: The vehicle speed is divided into 10 levels with 10 km / h as one level, and the speed range of 0-100 km / h is divided into 10 levels. The road surface grade is taken into 5 levels AE according to the road conditions, so as to traverse the actual situation of the vehicle driving on different road surface grades at each speed level.

[0104] ③ Signal decomposition: Perform variational modal decomposition on the suspension dynamic deflection and sprung mass acceleration signals, and extract characteristic components (BIMF components) respectively. There are many types of characteristic components, such as mean, root mean square value, root amplitude, variance, peak value and standard deviation. However, some statistical features are not sensitive to the time-frequency domain information caused by different road surface grades. At this time, this part of the statistical features should be eliminated, and those most obvious difference features due to different road surface grades should be retained. Therefore, the BIMF components after screening obviously contain road surface grade information.

[0105] ④ Signal feature calculation: The selected BIMF components containing road surface information are reconstructed to form an initial feature matrix, and the singular value entropy of the matrix is ​​calculated.

[0106] ⑤ Road surface grade classification: The calculated singular value entropy results are imported into ANFIS, and fuzzy rules are automatically established through machine learning to obtain the corresponding relationship between singular value entropy and road surface grade. Finally, road surface grade recognition can be realized according to the response signal, and the final road surface grade is output. At the same time, the information will also be uploaded to the cloud to correct the map.

[0107] In this embodiment, the confirmation method for pulse excitation such as pits and speed bumps is as follows:

[0108] The location of pulse excitations with obvious features such as speed bumps can be confirmed based on high-precision maps. After the front vehicle passes, this type of location information will be uploaded to the cloud to correct the map. At the same time, this information is also sent to the rear vehicle. The vehicle ECU of the rear vehicle will calculate the time it takes for the vehicle to arrive at the pulse excitation location based on the vehicle speed information provided by the low-orbit satellite and its own positioning information.

[0109] In this embodiment, vehicles rely on a communication system to transmit signals, sending real-time information such as the road surface grade, suspension parameters, and self-positioning to the vehicle behind them, while receiving similar information from adjacent vehicles in order to more comprehensively understand the road surface feedback status under local road conditions, so that the following vehicle can match the suspension control parameters.

[0110] The existing data communication method between paired vehicles generally uses WI-FI or Bluetooth mode, which has poor penetration and a short effective distance.

[0111] To address this shortcoming, the present invention can find a suitable base station between the two vehicles, such as a base station close to both vehicles, based on the positioning information provided by the combined positioning system in step 1, after determining the position of the paired vehicles and the distance between the two vehicles, and use the base station to transmit information. Figure 2As shown in the figure, in the high-precision map, the front vehicle passes a certain location, obtains the local point information and then sends a communication request to the rear vehicle through the base station. After the rear vehicle responds, the two vehicles establish a TCP connection. The front vehicle then sends a data packet of position coordinate information, suspension response parameters, and road surface information to the rear vehicle. At the same time, the fleet will also establish a TCP connection with the cloud server to update the real-time road section information and send the data packet to other vehicles following behind, completing the intelligent monitoring and control management of vehicles and roads on the entire road section.

[0112] Step 3: Beyond visual range preview control strategy: The present invention considers the control force that the corresponding suspension actuator should output under the two conditions of random road surface and pulse excitation. Under random road surface, the suspension actuator still matches the fuzzy control quantization factor according to the road surface grade of the preceding vehicle and outputs the fuzzy control force; under pulse excitation, the suspension actuator output becomes the LQG control force. Specifically, the beyond visual range preview control strategy is formulated based on the received road surface state, suspension parameters, and vehicle positioning information.

[0113] First, we should confirm that the control target of the active suspension system is to ensure vehicle smoothness and meet the requirements of handling stability. At this time, the vehicle should ensure:

[0114] ① The active suspension system actuator has output saturation, and the control force generated by the algorithm should be less than the output limit of the actuator:

[0115] F≤F max (10)

[0116] ② Due to the mechanical structure limitation of the suspension itself, the suspension dynamic deflection should be limited within a certain range. In order to ensure that it is suitable for most vehicles, try to design it according to the minimum dynamic deflection of most vehicles:

[0117] Z s -Z t ≤Z max (11)

[0118] ③ During the driving process, the dynamic load between the tire and the ground at any time should be less than the static load when the vehicle is stationary, to ensure the vehicle has stable handling, which can be:

[0119] K t (Z t -Z g )≤(M u +M t )g (12)

[0120] On this basis, the vehicle's beyond-visual-range preview control is designed. The entire vehicle suspension control process is as follows: Figure 5When the vehicle passes through the road, the front vehicle provides the rear vehicle with the grade information of the road it is about to pass through. The vehicle suspension controller switches the quantization factor parameters in time according to the road information ahead to reduce the working pressure of the controller.

[0121] If there is no pulse excitation on the road surface where the front vehicle passes, the vehicle suspension controller adopts fuzzy control, that is, by designing a set of fuzzy rules to deal with the entire control process, and only the quantization factor before the fuzzy control input changes with the road information.

[0122] The fuzzy control process is as follows:

[0123] ② In order to better follow the effect of the road grade recognition algorithm, the input of the controller is set as the difference e and the change rate ec between the suspension dynamic deflection in the suspension system and the ideal value.

[0124] ③ Select the basic domain and quantization factor of the input and output signals. The output signal can set the output upper limit of the actuator as its domain range. The input signal domain selects the response range of the suspension dynamic deflection of the C-grade road surface at a speed of 50km / h, and uses this as a standard to calculate the quantization factor value K under different road surfaces at different speeds. i .

[0125] ④ Design of membership function: The input variables are e and ec, and 7 linguistic fuzzy subsets {NB (negative large), NM (negative medium), NS (negative small), Z (zero), PS (positive small), PM (positive medium), PB (positive large)} are used. To ensure that the output of the algorithm can transition smoothly, the membership functions are all trimf-shaped.

[0126] ⑤ Design of fuzzy rules: According to the relationship between input and output, combined with the system's requirements for fuzzy controllers, fuzzy reasoning rules are designed. Therefore, the basic principles for designing fuzzy control rules can be concluded as follows:

[0127] (1) When the values ​​of e and ec are large and in the same direction, the system error is large and the error is still increasing. This requires the system to provide a large control amount to reduce its error and ensure the stability of the system.

[0128] (2) When the values ​​of e and ec are large and in opposite directions, this state indicates that the system error is large but decreasing. This requires the system to try to ensure that the control amount is not too large, to ensure that the system does not produce large oscillations, and to ensure that the output of the control force is stable;

[0129] (3) When the values ​​of e and ec are small, the system begins to approach stability. When the input value is in the same direction, a very small output force should be selected. When the input value is in the opposite direction, the output force can be kept at zero to prevent the system from oscillating.

[0130] The fuzzy control rule table can be obtained:

[0131]

[0132]

[0133] If there is pulse excitation, the LQG control method containing preview information is as follows:

[0134] When the vehicle is subjected to pulse excitation, in order to better cope with such instantaneous large impacts and ensure the integrity of the goods, the LQG algorithm with preview information can achieve more accurate force tracking control. The specific process is as follows Figure 6 shown.

[0135] The calculation process of the LQG algorithm with preview information is as follows:

[0136] Rewrite equations (5)-(9) into state equations:

[0137]

[0138] Among them, x is the state variable, is the first-order derivative of the state variable, y is the output variable, u is the control input, w is the noise input, and A, B, E, C, and D are the coefficient matrices of each item.

[0139] Let x1 = Z uf 、x2=Z sf 、x3=Z ur 、x4=Z sr , the road surface input x passed by the front wheel 01 Provided by the vehicle in front, since the input at the rear wheels is only a time lag compared to the input at the front wheels (wheelbase / vehicle speed), so the input of the rear wheel can be considered as x 02 =x 01 (t-τ).

[0140] The state variables are:

[0141]

[0142] The output variables are:

[0143]

[0144] The control input is:

[0145] u=[F f F r ]

[0146] The noise input is:

[0147] w=[x 01 x 02 ]

[0148] make

[0149] Coefficient matrix:

[0150]

[0151] The vehicle body acceleration and front suspension dynamic deflection are used to evaluate the driver's riding comfort, the rear suspension vertical acceleration and rear suspension dynamic deflection are used to evaluate the cargo integrity, and the front and rear tire dynamic loads (since the dynamic load is proportional to the tire dynamic stroke, the tire dynamic displacement is used instead) are used to evaluate the road friendliness. Considering the limitation of control energy, the performance index functional of the optimal controller of the active suspension is defined as:

[0152]

[0153] In the formula, q1, q2, q3, q4, q5, q6, r1 and r2 are weighted coefficients of each indicator. Then the indicator functional can be transformed into:

[0154]

[0155] In the formula, q is a vector composed of q1, q2, q3, q4, q5, and q6, denoted as q=diag(q1,q2,q3,q4,q5,q6); r is a vector composed of r1 and r2, denoted as r=diag(r1,r2); Y is the output vector matrix, U is the optimal control vector of the system, and t is the time variable of the time domain signal.

[0156] Introducing the output variable, the indicator functional is further transformed into:

[0157]

[0158] In the formula, X is the state variable of the system, Q is the weight coefficient matrix of the state variable, denoted as Q = C T qC; N is the cross-term weight coefficient matrix; R is the control variable weight coefficient matrix, denoted by R = r + D T qD;N=C T qD

[0159] At this point, the optimal control vector of the system can be obtained as:

[0160] U=-KX=-R -1 (N T +B T L)X (17)

[0161] Where K is the feedback gain matrix of LQG control, B is the coefficient matrix of system control, and L is the solution of the Riccati equation:

[0162] LA+AT L-LBR -1 B T L+Q=0 (18)

[0163] Among them, A is the system state coefficient matrix.

[0164] The Laplace transfer function and its second-order Pade approximation are used to express the relationship between the road inputs at the front and rear wheels:

[0165]

[0166] Where s is a complex variable in Laplace space, and the coefficient coefficient Coefficient a2 = 1;

[0167] Take an additional state vector η = [η1η2] and transform the above equation (19) into the state equation form:

[0168]

[0169] In the formula,

[0170] Combining equation (20) with the system state space equation (13), the system state equation combined with preview control and with an additional state vector can be obtained as:

[0171]

[0172] In the formula, the matrix matrix

[0173] The new system state equation is combined with the previous performance index functional equation J to solve the Riccati equation, and the optimal control gain matrix of the wheelbase preview information is combined to obtain the optimal control force of the wheelbase preview:

[0174] U p (t) = K p [X(t)+η(t)] (22)

[0175] Among them, K p is the optimal control gain matrix combined with the wheelbase preview information.

[0176] Through the above implementation modes, the present invention can realize road condition recognition beyond visual range and perform precise preview control on the active suspension, thereby effectively improving the driving comfort of the vehicle under complex road conditions.

[0177] The above embodiments are only used to illustrate the design ideas and features of the present invention, and their purpose is to enable those skilled in the art to understand the content of the present invention and implement it accordingly. The protection scope of the present invention is not limited to the above embodiments. Therefore, any equivalent changes or modifications made based on the principles and design ideas disclosed by the present invention are within the protection scope of the present invention.

Claims

1. A beyond-visual-range active suspension preview control method based on high-precision positioning road state recognition, characterized in that: The steps include: Step 1: Based on the combined positioning system of low-orbit satellite positioning, high-precision map and inertial navigation system, the vehicle is positioned to obtain the vehicle's location information; Step 2: Based on the position information of the vehicle, a road surface state is identified, where the road surface state includes a road surface grade and a pulse excitation; It also establishes signal interaction between vehicles, and realizes interactive transmission of road conditions, suspension parameters, and vehicle positioning information between the vehicle and surrounding vehicles; Step 3: Based on the received road surface conditions, suspension parameters, and vehicle positioning information, a beyond-visual-range preview control strategy is formulated; if there is no pulse excitation on the road section, the vehicle suspension controller adopts fuzzy control; if there is a pulse excitation, the vehicle suspension controller adopts LQG control containing preview information.

2. The beyond-visual-range active suspension preview control method based on high-precision positioning road state recognition according to claim 1 is characterized in that: In the process of positioning the vehicle, the combined positioning system uses the inertial navigation system to obtain the vehicle mileage, the high-precision map provides candidate roads based on the vehicle mileage, and the low-orbit satellite provides the distance from the vehicle to the candidate road and the angle between the vehicle heading and the direction of the candidate road. The real positioning information of the vehicle is obtained in the map through fusion calculation.

3. The beyond-visual-range active suspension preview control method based on high-precision positioning road state recognition according to claim 2 is characterized in that: The method for determining the real positioning information of the vehicle by fusion calculation is as follows: after obtaining the distance d from the vehicle to the candidate road i and the angle ξ between the vehicle heading and the candidate road direction i Finally, the distribution probability corresponding to each candidate road is calculated using Gaussian distribution; the candidate road corresponding to the maximum distribution probability is selected, and the vehicle road and position are located using the weighted matching value.

4. The beyond-visual-range active suspension preview control method based on high-precision positioning road state recognition according to claim 1, characterized in that: The road surface grade identification method is: ① Obtaining suspension response signals, including: collecting the vehicle's vertical acceleration signal through an acceleration sensor, solving the suspension dynamic deflection based on the vehicle dynamics equation, and obtaining the vehicle's body speed through a low-orbit satellite; ② Preprocessing of suspension response signals: The vehicle speed is divided into grades, and the road surface grade is graded according to the road conditions, so as to traverse the conditions of the vehicle driving on different road surfaces at each speed gear and obtain the vehicle response; ③Signal decomposition: Perform variational modal decomposition on the suspension dynamic deflection and sprung mass acceleration signals in the vehicle response signal, and extract characteristic components respectively; and select characteristic components that can obviously contain road surface grade information, so that the characteristic components can obviously contain road surface information; ④ Signal feature calculation: reconstruct the selected characteristic components containing road surface information to form an initial feature matrix, and calculate the singular value entropy of the matrix; ⑤ Road surface grade classification: Rely on machine learning to automatically establish fuzzy rules, obtain the corresponding relationship between singular value entropy and road surface grade, and realize road surface grade identification based on the response signal.

5. The beyond-visual-range active suspension preview control method based on high-precision positioning road state recognition according to claim 1, characterized in that: Method to confirm pulse excitation: The location where the pulse excitation is generated on the road surface is confirmed through a high-precision map. The vehicle in front that passes the location uploads the location information to the cloud to correct the map. At the same time, the location information is synchronously sent to the following vehicle. The vehicle ECU of the following vehicle calculates the time it takes for the vehicle to arrive at the pulse excitation location based on the vehicle speed information provided by the low-orbit satellite and its own positioning information.

6. The beyond-visual-range active suspension preview control method based on high-precision positioning road state recognition according to claim 1, characterized in that: The method of establishing signal interaction between vehicles is: Based on the positioning information provided by the combined positioning system, after determining the position of the paired vehicles and the distance between the two vehicles, the base station is found in the area between the two vehicles and the base station is used to transmit information.

7. The beyond-visual-range active suspension preview control method based on high-precision positioning road state recognition according to claim 1, characterized in that: The fuzzy control process is as follows: 1) The input of the controller is defined as the difference e and the change rate ec between the suspension dynamic deflection and the ideal value in the suspension system; 2) Select the basic domain and quantization factor of the input signal and output signal respectively. The output signal can select the output upper limit of the suspension actuator as its domain range. The input signal domain selects the response range of the suspension dynamic deflection on a C-grade road surface at a speed of 50km / h, and uses this as a standard to calculate the quantization factor value K under different road surfaces at different speeds. i ; 3) Design membership function: The input variables are e and ec, and 7 linguistic fuzzy subsets {NB (negative large), NM (negative medium), NS (negative small), Z (zero), PS (positive small), PM (positive medium), PB (positive large)} are used. The membership functions are all trimf-shaped; 4) Design fuzzy rules: (1) When the values ​​of e and ec are large and in the same direction, the system error is large and the error is still increasing. The system needs to provide a large control amount to reduce the error and ensure the stability of the system. (2) When the values ​​of e and ec are large and in opposite directions, it means that the system error is large but decreasing. The system needs to try to ensure that the control amount is not too large, that the system does not produce large oscillations, and that the output of the control force is stable. (3) When the values ​​of e and ec are small, the system begins to approach stability. When the input value is in the same direction, a very small output force should be selected. When the input value is in the opposite direction, the output force can be kept at zero to prevent the system from oscillating.

8. The beyond-visual-range active suspension preview control method based on high-precision positioning road state recognition according to claim 1 is characterized in that: The LQG control method is as follows: Construct the vehicle dynamics equation and rewrite it into a state equation, recorded as: y=Cx+Du Let x1 = Z uf 、x2 = Z sf 、x3 = Z ur 、x4 = Z sr , so The state variables are: The output variables are: The control input is: The noise input is: w = [x 01 x 02 ]; Among them, A, B, C, D, E are coefficient matrices, x 01 is the road surface input passed by the front wheel, x 02 is the road surface input of the rear wheels; Z uf , Z ur is the vertical displacement of the front and rear tires of the vehicle; Z sf , Z sr F is the vertical displacement of the front and rear body of the vehicle; f 、F r Output force for its front and rear active suspension actuators; The vehicle body acceleration and front suspension dynamic deflection are used to evaluate the driver's riding comfort, the rear suspension vertical acceleration and rear suspension dynamic deflection are used to evaluate the cargo integrity, and the front and rear tire dynamic loads are used to evaluate the road friendliness. Considering the limitation of control energy, the performance index functional of the active suspension optimal controller is defined as: Among them, q1, q2, q3, q4, q5, q6, r1 and r2 are weighted coefficients of each indicator; is the vertical acceleration of the vehicle's center of mass, and t is the time variable of the time domain signal. Introducing the output variable, the indicator functional is further transformed into: In the formula, X is the state variable of the system, Q is the weight coefficient matrix of the state variable, N is the cross-term weight coefficient matrix; R is the control variable weight coefficient matrix; U is the control vector of the system. The optimal control vector of the system is obtained as: U=-KX=-R -1 (N T +B T L)X Where K is the feedback gain matrix of LQG control, B is the coefficient matrix of system control, and L is the solution of the Riccati equation. After combining the additional state vector with the system state space equation, the system state equation combined with preview control and with the additional state vector is: In the formula, is the first-order derivative of the system’s state variables, is the first-order derivative of the additional state vector, A is the coefficient matrix of the system, F is the coefficient matrix of the additional state vector, A η , D η 、E η , B η are the coefficient matrices of the system after adding the state vector, respectively a0 and a1 are coefficients, respectively denoted as coefficients coefficient The above system state equation is combined with the previous performance index functional equation J to solve the Riccati equation, and combined with the optimal control gain matrix of the wheelbase preview information, the optimal control force of the wheelbase preview can be obtained: U p (t)=K p [X(t)+η(t)] Among them, K p is the optimal control gain matrix combined with the wheelbase preview information, X(t) is the system state variable in the time domain, and η(t) is the additional state vector in the time domain.

9. An electronic device, characterized in that: The invention comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the above-mentioned beyond-visual-range active suspension preview control method based on high-precision positioning road state recognition is implemented.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by the processor, the above-mentioned beyond-visual-range active suspension preview control method based on high-precision positioning road state recognition is implemented.

Images