A novel LQG control method

Abstract

The present invention provides a novel LQG control method, including: Step 1: Establish a seven-degree-of-freedom active suspension simulation model for the whole vehicle; Step 2: Establish thresholds for the novel LQG control method of each performance index of the suspension; Step 3: Construct an optimized fitness function for the novel LQG control method; Step 4: Optimize to obtain the state weighting coefficients of the novel LQG control; Step 5: Calculate the optimal feedback matrix K of the novel LQG control; Step 6: Finally construct the novel LQG control method according to the optimal feedback matrix K, and its output control force U. This method uses the root mean square values of the body weighted acceleration, suspension dynamic deflection, and tire dynamic load as indicators, and quantifies the evaluation method of the novel LQG control effect of ride comfort and safety. The LQG control parameters corresponding to different road surface grades under this evaluation method are determined by genetic algorithm optimization. Simulation verifies that the novel LQG control method has better practicability and optimized control effect.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle control, and particularly to a novel LQG control method. Background Art

[0002] Since active suspensions can provide active control forces and can perform adaptive control according to the current driving conditions of the vehicle and actively adjust the suspension state, they have received extensive attention. The key to active suspension control lies in selecting an appropriate control strategy. Therefore, the current active suspension control strategy has become the focus of research by many scholars. Stochastic linear optimal control (LQG) has the characteristics of linearization, simple design, and strong adaptability. Since it was proposed, articles on the stochastic linear optimal control of vehicle active suspensions have been continuously published. The key to designing a stochastic linear optimal control strategy lies in the selection of state weighting coefficients. In reference [1] , the analytic hierarchy process was used to determine the state weighting coefficients. In references [2] and [3] , the genetic algorithm and the genetic particle swarm algorithm were respectively used to determine the state weighting parameters, and good control effects were achieved. However, the current methods for determining state weighting parameters are mostly qualitative analysis or simply based on minimizing the numerical value of the evaluation index, and do not consider the different requirements of the vehicle for each suspension performance index under different road surface grades, so that the control method cannot meet the actual needs of the current road surface suspension to the greatest extent and achieve the optimal control effect.

[0003] [1] Luo Xinyuan, Yang Shiwen. Design of LQG Controller for Vehicle Active Suspension Based on AHP [J]. Control Theory & Applications, 2004, 21(1): 139 - 144.

[0004] [2] Meng Jie, Zhang Kai, Jiao Hongyu. Design of LQG Controller for Automotive Active Suspension Optimized by Genetic Algorithm [J]. Mechanical Science and Technology, 2013, 32(6): 914 - 918.

[0005] [3] Chen Shuang, Zong Changfu. Genetic Particle Swarm LQG Control Method for Vehicle Active Suspension [J]. Automotive Engineering, 2015, 37(2): 189 - 193.

[0006] [4] Chen Ying. Design of LQG Controller for Vehicle Suspension [D]. Xi'an University of Technology, 2017.

[0007] [5] Wu Wangsheng. Research on Semi - active Suspension Based on Firefly LQG Control Algorithm [D]. Anhui Polytechnic University, 2017.

[0008] [6] Chen Shian, Zu Guanghao. Taylor Series-LQG Time Delay Compensation Control Method for Magnetorheological Semi-Active Suspension [J]. Journal of Vibration and Shock, 2017, 3(8): 190-196.

[0009] [7] Zhu Tianjun, Zong Changfu, Yang Dejun, et al. Subjective Evaluation Method for Automobile Ride Comfort [J]. Automobile Technology, 2008(3): 8-11.

[0010] [8] Yu Zhisheng. Automobile Theory. Fifth Edition [M]. Beijing: China Machine Press, 2009.

[0011] [9] Thompson A G. An Active Suspension with Optimal Linear State Feedback [J]. Vehicle System Dynamics, 1976, 5(4): 187-203.

[0012]

[10] Chen Shian, Qiu Feng, He Ren, Lu Senlin. A Method for Determining the Weighting Coefficients of Vehicle Suspension LQG Control [J]. Journal of Vibration and Shock, 2008, 27(2): 65-68.

[0013]

[11] Zhao Heng, Lu Shifu. Time Domain Model of Road Input to Four-Wheel Vehicles [J]. Automotive Engineering, 1999, 21(2): 112-117. Summary of the Invention

[0014] The object of the present invention is to solve the defects existing in the above-mentioned prior art and provide a new LQG control method.

[0015] A new LQG control method includes the following steps:

[0016] Step 1: Establish a seven-degree-of-freedom active suspension simulation model for the whole vehicle;

[0017] Step 2: Establish the thresholds of the LQG control method for each performance index of the suspension; the indexes include: tire dynamic load, suspension dynamic deflection, and body weighted acceleration;

[0018] Step 3: Based on the thresholds of each performance index obtained in Step 2, construct an optimization fitness function for the LQG control method using the analytic hierarchy process; when constructing, when a single index exceeds the threshold, it should be optimized first, and when multiple indexes exceed the threshold, optimize according to the set optimization priority weights;

[0019] Step 4: Based on the seven-degree-of-freedom simulation model of the whole vehicle established in Step 1, select the state weighting coefficients of the LQG control through the genetic algorithm and the optimized fitness function of the new LQG control constructed in Step 3;

[0020] Step 5: Calculate the optimal feedback matrix K of the new LQG control according to the state weighting coefficient;

[0021] Step 6: Finally, construct the new LQG control method according to the optimal feedback matrix K, and its output control force U is equal to -X(t)*K; X(t) is the feedback state variable at any time.

[0022] Furthermore, for the method described above, the method for establishing the seven-degree-of-freedom active suspension simulation model of the whole vehicle in Step 1 is as follows:

[0023] According to the vehicle body motion differential equation

[0024]

[0025]

[0026]

[0027] and the unsprung mass m i Differential equation

[0028]

[0029] Establish a seven-degree-of-freedom active suspension simulation model of the whole vehicle;

[0030] Among them, M and m i are the sprung mass and the unsprung mass of the corresponding suspension respectively, z i2 and z i1 are the displacements of the sprung mass and the unsprung mass of the corresponding suspension respectively, z, θ, ψ are the vehicle body displacement, vehicle body pitch angle, and vehicle body roll angle respectively, k i and k t are the spring elastic stiffness and tire elastic stiffness of the corresponding suspension respectively, c i is the actuator damping of the corresponding suspension, U i is the actuator active control force of the corresponding suspension, I y 、I x are the moments of inertia of the vehicle body about the Y-axis and X-axis respectively, a and b are the distances from the front axle and the rear axle to the vehicle body mass center respectively, l f 、l r are the wheelbase of the front axle and the rear axle respectively, q i is the displacement input of the road surface roughness to the corresponding suspension.

[0031] Furthermore, for the method described above, Step 2 includes the following steps:

[0032] Step 21: When the tire dynamic load is too large, it has a certain impact on both braking and handling stability. Based on Principle: When the root mean square value F of the dynamic load d is greater than 1 / 3 of the static load F s , there is a possibility that the tire will jump off the ground, seriously deteriorating the driving safety. Therefore, F d / (F s / 3) is set as the threshold value of the tire dynamic load;

[0033] Step 22: When the dynamic deflection of the suspension is too large, it will hit the suspension limit block, which has a certain impact on both ride comfort and driving safety. Based on the principle, when the root mean square value f of the dynamic deflection of the suspension is greater than 1 / 3 of the suspension limit stroke S, there is a possibility of hitting the limit block, causing impact and noise. Therefore, f / (S / 3) is set as the threshold value of the dynamic deflection of the suspension;

[0034] Step 23: The vehicle body acceleration affects ride comfort. According to the evaluation standard of ISO2631-1, its weighted root mean square acceleration value is related to people's subjective feelings, and 1.25 is set as the threshold value of the weighted acceleration of the vehicle body.

[0035] Furthermore, for the method described above, the construction method of the optimized fitness function of the LQG control method in step 3 is as follows:

[0036] Step 41: Determine the weight coefficient ratio of the vehicle body acceleration, tire dynamic load, and suspension dynamic deflection through the analytic hierarchy process to clarify the optimization priority of each performance parameter. When it does not exceed the threshold value, its weight coefficient ratio is set to 5:1:0; when it exceeds the threshold value, its weight coefficient ratio is set to 7:1;1;

[0037] Step 42: Design the penalty coefficient ξ = 10 to improve the optimization priority when the index exceeds the threshold value;

[0038] Step 43: According to 41, 42 and step 2, construct the optimized fitness function of each performance index:

[0039] According to people's subjective feelings, construct the fitness function R1 of the vehicle body acceleration

[0040]

[0041] Construct the fitness function R2 of the tire dynamic load

[0042]

[0043] Construct the fitness function R3 of the suspension dynamic deflection

[0044]

[0045] Then, the total fitness function is \(R = R_1+R_2 + R_3\).

[0046] Furthermore, for the method as described above, the construction method of the optimal feedback matrix \(K\) of the LQG control in step 5 is as follows:

[0047] Solve it through the lqr function in Matlab to calculate the optimal feedback matrix \(K\).

[0048] Beneficial effects:

[0049] The novel LQG control strategy provided by the present invention can comprehensively consider the influence of the performance of each suspension on driving under the current road surface grade, effectively and automatically select the optimization priority of each performance under the current road surface grade, and has better applicability. The present invention determines the corresponding LQG control parameters under different road surface grades. Taking the passive suspension and the traditional LQG control strategy as comparison objects, simulation verifies that the novel LQG control method has better practicability and optimal control effect. Description of the drawings

[0050] Figure 1 is the 7-degree-of-freedom suspension model of the whole vehicle;

[0051] Figure 2(a) is a comparison diagram of the dynamic tire displacement of the left front suspension of the passive suspension, the traditional LQG and the novel LQG-controlled active suspension under the D-level road surface;

[0052] Figure 2(b) is a comparison diagram of the dynamic deflection of the left front suspension of the passive suspension, the traditional LQG and the novel LQG-controlled active suspension under the D-level road surface;

[0053] Figure 2(c) is a comparison diagram of the vertical acceleration of the vehicle body of the passive suspension, the traditional LQG and the novel LQG-controlled active suspension under the D-level road surface;

[0054] Figure 2(d) is a comparison diagram of the pitch angular velocity of the vehicle body of the passive suspension, the traditional LQG and the novel LQG-controlled active suspension under the D-level road surface;

[0055] Figure 2(e) is a comparison diagram of the roll acceleration of the vehicle body of the passive suspension, the traditional LQG and the novel LQG-controlled active suspension under the D-level road surface;

[0056] Figure 3(a) is a comparison diagram of the dynamic displacement of the left front wheel of the passive suspension, the traditional LQG and the novel LQG-controlled active suspension under the B-level road surface;

[0057] Figure 3(b) is a comparison diagram of the dynamic deflection of the left front suspension of the passive suspension, the traditional LQG and the novel LQG-controlled active suspension under the B-level road surface;

[0058] Figure 3(c) is a comparison chart of the body vertical acceleration of the passive suspension, the traditional LQG, and the new LQG-controlled active suspension under the B-class road surface;

[0059] Figure 3(d) is a comparison chart of the body pitch acceleration of the passive suspension, the traditional LQG, and the new LQG-controlled active suspension under the B-class road surface;

[0060] Figure 3(e) is a comparison chart of the body roll acceleration of the passive suspension, the traditional LQG, and the new LQG-controlled active suspension under the B-class road surface. Detailed implementation manner

[0061] To make the objectives, technical solutions, and advantages of the present invention clearer, the technical solutions in the present invention will be clearly and completely described below. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without making creative efforts belong to the scope of protection of the present invention.

[0062] Since the control method involved in the present invention requires a relatively complete suspension working condition, and the vehicle seven-degree-of-freedom model can comprehensively reflect the suspension working condition, a vehicle 7-degree-of-freedom active suspension model is established, as Figure 1 shown.

[0063] The differential equation of the body motion is as follows:

[0064]

[0065]

[0066]

[0067] The unsprung mass m i The differential equation is as follows:

[0068]

[0069] Among them, M and m i are the sprung mass and the unsprung mass of the corresponding suspension respectively, z i2 and z i1 are the displacements of the sprung mass and the unsprung mass of the corresponding suspension respectively, z, θ, ψ are the body displacement, body pitch angle, and body roll angle respectively, k i and k t are the spring elastic stiffness and the tire elastic stiffness of the corresponding suspension respectively, c i is the actuator damping of the corresponding suspension, U i is the active control force of the actuator of the corresponding suspension, I y 、I xare the moments of inertia of the vehicle body about the Y-axis and X-axis respectively, a and b are the distances from the front axle and the rear axle to the vehicle body's center of mass, and l f and l r are the wheelbase lengths of the front axle and the rear axle respectively, q i is the displacement input of the road surface unevenness to the corresponding suspension.

[0070] The present invention uses the filtered white noise method to simulate the road surface unevenness input. The literature

[11] points out that there is a correlation between the road surface unevenness inputs between the left and right wheels on the same axle, and they interfere with each other. Then, the state equation of the four-wheel coherent road excitation input is:

[0071]

[0072]

[0073]

[0074] P = [b -b Db -Db O 2×1 ′

[0075]

[0076] a = 2πf0u

[0077] where ξ = [ξ1 ξ2]′ is the intermediate variable, u is the vehicle speed (m / s); f0 is the lower cut-off frequency, equal to 0.011u; n0 is the spatial reference frequency, taken as 0.1; w is the Gaussian white noise; G q (n0) is the road surface unevenness coefficient, l is the wheelbase length between the front and rear axles. Referring to the literature, a0 = 3.1851, a1 = 0.2063, a2 = 0.0108, b0 = 3.223, b1 = 0.59, b2 = 0.0327 can be obtained.

[0078] Based on equations (1) and (2), the following state space is established

[0079]

[0080] where X is the system state vector

[0081] X = [x1 x2 x3 x4 x5 x6 x7 x8 x9 x 10 x 11 x 12 x 13 x 14 x 15 ′

[0082] x1 = z 11 -q1 x3 = z31 -q3

[0083] x2 = z 21 -q2 x4 = z 41 -q4 x5 = z 12 -z 11

[0084] x6 = z 22 -z2 x7 = z 32 -z 31 x8 = z 42 -z 41

[0085]

[0086] x 15 = z 41

[0087] U is the input matrix, which is the active control force [f1 f2 f3 f4]′ in this model

[0088] W is the Gaussian noise input matrix, that is Design of LQG Controller for a 7-Degree-of-Freedom Full-Vehicle Active Suspension

[0089] Since the dynamic tire displacement affects the driving safety of the vehicle, the body acceleration affects the ride comfort, and the excessive dynamic deflection of the suspension will also deteriorate the safety and comfort. When designing the LQG controller, it is necessary to comprehensively consider the optimal control of these three factors. Based on this, the control error J of the LQG controller for the 7-degree-of-freedom full-vehicle active suspension is designed as follows:

[0090]

[0091] It is rewritten in the form of a quadratic function integral as follows

[0092]

[0093] The relevant matrices are obtained according to the 7-degree-of-freedom full-vehicle suspension state space given by formula (6).

[0094] When the input and state weighting coefficients of the model are determined, the optimal control feedback gain matrix K can be given by the Riccati equation, and its form is:

[0095] AK + KA′ + Q - KBR -1 B′K + FWF′ = 0 (9)

[0096] According to the control law of LQG control theory, from the feedback state variable X(t) at any time, the optimal control force U of the actuator at time t can be obtained.

[0097] The state weighting coefficient directly affects the control effect of LQG. Currently, the methods for determining the LQG state weighting coefficient are mainly divided into two types: the analytic hierarchy process and the optimization algorithm. Since the former has subjective factors, it is difficult to achieve the optimal control effect. Therefore, the latter is currently mostly used to determine the state weighting coefficient. For example, in the literature [5] the genetic algorithm is adopted, and in the literature [6] the firefly swarm algorithm is adopted for parameter optimization.

[0098] Currently, the fitness function R of the optimization algorithm is mostly

[0099]

[0100] Among them, are the body accelerations of the suspension and the passive suspension under LQG control respectively; are the body pitch angular accelerations of the suspension and the passive suspension under LQG control respectively; are the body roll angular accelerations of the suspension and the passive suspension under LQG control respectively; f 1i 、f 0i are the tire dynamic loads of the suspension and the passive suspension under LQG control respectively; d 1i 、 d 0i are the suspension dynamic deflections of the suspension and the passive suspension under LQG control respectively. It can be seen that the basic idea of its control strategy is to pursue the maximization of numerical optimization. However, this fitness function does not consider the different priorities of the optimization of the three parameters under different road grades, resulting in the control applicability not reaching the optimal.

[0101] Therefore, the present invention designs an evaluation method for the LQG control effect that quantifies ride comfort and safety with the root mean square values of body acceleration, suspension dynamic deflection, and tire dynamic load as indicators:

[0102] Body acceleration mainly affects ride comfort. According to the evaluation standard of ISO2631-1, the weighted acceleration root mean square value has the relationship with people's subjective feelings as shown in Table 1:

[0103] Table 1 Relationship between weighted acceleration root mean square value and people's subjective feelings

[0104]

[0105] People's subjective feelings are affected by personal factors, and there is an overlap phenomenon of weighted acceleration values in different feeling intervals. To ensure the control effect, we select the numerical intersection point between this interval and the next interval as the upper limit of this interval.

[0106] It is difficult to directly compare the subjective feeling of ride comfort based on the root mean square value of body weighted acceleration with the tire dynamic load and suspension dynamic deflection, and there is incommensurability. Therefore, in this invention, taking the passive suspension as a reference, through genetic algorithm optimization, when the root mean square value of body weighted acceleration is at the upper limit of each feeling interval, the root mean square value of the tire dynamic load of the passive suspension is obtained. Taking this root mean square value of the dynamic load as the basic weight value of this feeling interval, when the root mean square value of the tire dynamic load is near the threshold, the root mean square value of the weighted acceleration of the passive suspension is used as the threshold of the root mean square value of the body weighted acceleration.

[0107] The tire dynamic load mainly affects driving safety. Excessive dynamic load has a certain impact on both braking and handling stability. Based on the principle, when the root mean square value of the dynamic load F d is greater than 1 / 3 of the static load F s , there is a possibility that the tire will leave the ground, seriously deteriorating driving safety.

[0108] The suspension dynamic deflection has a certain impact on both ride comfort and driving safety. Based on the principle, when the root mean square value of the suspension dynamic deflection f is greater than 1 / 3 of the suspension limit travel S, there is a possibility of hitting the limit block, causing impact and noise.

[0109] By referring to the relevant literature of AHP

[10] , in the case of better working conditions (A and B grade roads), the subjective weight coefficients of body acceleration, tire dynamic load, and suspension dynamic deflection are mostly set to 5, 1, 0; in the case of worse working conditions (D grade roads), the subjective weight coefficients of body acceleration, tire dynamic load, and suspension dynamic deflection are mostly set to 7, 1, 1. This invention takes this subjective weight coefficient as the priority weight coefficient of LQG control for the three indicators under the same suspension working conditions.

[0110] The control idea of this invention is that when a single indicator exceeds the threshold, it should be optimized first, and when multiple indicators exceed the threshold, they should be optimized according to the set optimization priority weight.

[0111] Design the penalty coefficient ξ = 10 to improve the optimization priority when the indicator exceeds the threshold.

[0112] Based on the above idea, the following fitness function of LQG control is designed

[0113] (1) According to the subjective feeling of the human body, construct the body acceleration fitness function R1

[0114]

[0115] (2) Construct the tire dynamic load fitness function R2

[0116]

[0117] (3) Construct the suspension dynamic deflection fitness function R3

[0118]

[0119] Then the total fitness function is R = R1 + R2 + R3

[0120] The new fitness function can determine different optimization priorities according to the deterioration degrees of vehicle body acceleration, suspension dynamic deflection, and tire dynamic load, improving its control effect and control applicability.

[0121] Example simulation

[0122] To verify the effectiveness of the new LQG control strategy, a passive suspension, an active suspension based on the traditional LQG control strategy, and an active suspension based on the new LQG control strategy were selected for performance comparison and analysis.

[0123] The selected suspension parameters are shown in Table 3:

[0124] Table 3 Suspension parameters

[0125]

[0126] Simulation was carried out under the conditions of D-class road surface, vehicle speed of 10 m / s, and B-class road surface, vehicle speed of 30 m / s.

[0127] The genetic algorithm (GA) was used to optimize the state weighting coefficients of the traditional LQG control strategy and the new LQG control strategy. The corresponding state weighting coefficients are shown in Table 4:

[0128] Table 4 State weighting coefficients for B-class road surface

[0129]

[0130] Table 5 State weighting coefficients for D-class road surface

[0131]

[0132] Using the previously established state space equation, [K, S, E] = lqr(A, B, Q, R, N) in Matlab can be called to calculate the optimal feedback matrix K. Substituting the calculated quantized total optimal feedback matrix into the Matlab / Simulink simulation block diagrams of the active suspension and the passive suspension, the simulation curves of each performance index are obtained. Figures 2(a)-(e) and Figures 3(a)-(e) are respectively the comparison diagrams of each performance index of the passive suspension, the traditional LQG, and the new LQG control active suspension under D-class and B-class road surfaces.

[0133] The root mean square values of the performance evaluation indexes of the three suspension systems are shown in Table 5

[0134] Table 6 Root mean square values of the performance evaluation indexes of the three suspension systems

[0135]

[0136]

[0137] From Figures 2(a)-(e) and Table 6, it can be analyzed that:

[0138] (1) On a Class B road surface, compared with the traditional LQG control, the body acceleration of the new LQG control deteriorates (2.5%), but it is still within the agreed subjective comfort range, and the deterioration of ride comfort can be ignored. However, the dynamic tire displacement decreases by 14.2% compared with the traditional LQG control, effectively improving driving safety.

[0139] (2) On a Class D road surface, due to the poor road conditions and all three indexes being in a severe working condition, compared with the traditional LQG control, when the body acceleration of the new LQG control deteriorates slightly (3.06%), it ensures that the dynamic deflection of the suspension is below the threshold value (0.0266), preventing the occurrence of hitting the stop block and improving ride comfort and driving safety.

[0140] Through the above analysis, it can be seen that compared with the traditional LQG control, the new LQG control strategy provided by the present invention can comprehensively consider the influence of the performance of each suspension on driving under the current road surface grade, effectively automatically select the optimization priority of each performance under the current road surface grade, and has better applicability.

[0141] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A novel LQG control method, characterized in that, It includes the following steps: Step 1: Establish a seven-degree-of-freedom active suspension simulation model for the whole vehicle; Step 2: Establish the thresholds of the LQG control method for each performance index of the suspension; the indexes include: tire dynamic load, suspension dynamic deflection, and body weighted acceleration; Step 3: Based on the thresholds of each performance index obtained in Step 2, construct an optimization fitness function for the LQG control method using the analytic hierarchy process; when constructing, when a single index exceeds the threshold, it should be optimized first, and when multiple indexes exceed the threshold, optimize according to the set optimization priority weights; Step 4: Based on the seven-degree-of-freedom simulation model of the whole vehicle established in Step 1, select the state weighting coefficient of the LQG control through the genetic algorithm and the optimized fitness function of the new LQG control constructed in Step 3; Step 5: Calculate the optimal feedback matrix K of the new LQG control according to the state weighting coefficient; Step 6: Finally, construct a new LQG control method according to the optimal feedback matrix K, and its output control force U = [U1; U2; U3; U4] is equal to -X(T)*K; X(T) is the feedback state variable at any time; The method for establishing the seven-degree-of-freedom active suspension simulation model of the whole vehicle in Step 1 is as follows: According to the body motion differential equation and the unsprung mass M I Differential equation Establish a seven-degree-of-freedom active suspension simulation model for the whole vehicle; where M and M I are the sprung mass and the unsprung mass of the corresponding suspension respectively, Z I2 and Z I1 are the displacements of the sprung mass and the unsprung mass of the corresponding suspension respectively, Z, Θ, and Ψ are the body displacement, the body pitch angle, and the body roll angle respectively, K I and K T are the spring elastic stiffness and the tire elastic stiffness of the corresponding suspension respectively, C I is the actuator damping of the corresponding suspension, U I is the active control force of the actuator of the corresponding suspension, I Y and I X are the moments of inertia of the body about the Y-axis and the X-axis respectively, A and B are the distances from the front axle and the rear axle to the center of mass of the body respectively, L F and L R are the wheelbase of the front axle and the rear axle respectively, Q I is the displacement input of the road surface unevenness to the corresponding suspension.

2. The method according to claim 1, wherein Step 2 includes the following steps: Step 21: When the dynamic load of the tire is too large, it has an impact on both braking and handling stability. Based on the principle, when the root mean square value of the dynamic load F D is greater than 1 / 3 of the static load F S , there is a possibility that the tire will jump off the ground, seriously deteriorating the driving safety. Therefore, set F D / (F S / 3) as the threshold value of the tire dynamic load; Step 22: When the dynamic deflection of the suspension is too large, it will hit the suspension stop block, which affects both ride comfort and driving safety. Based on the principle, when the root mean square value F of the dynamic deflection of the suspension is greater than 1 / 3 of the suspension stop travel S, there is a situation of hitting the stop block, causing impact and noise. Therefore, set F / (S / 3) as the threshold value of the dynamic deflection of the suspension; Step 23: The vehicle body acceleration affects ride comfort. According to the evaluation criteria of ISO2631-1, the root mean square value of its weighted acceleration is related to people's subjective feelings, and 1.25 is set as the threshold value of the vehicle body weighted acceleration.

3. The method according to claim 2, wherein The method for constructing the optimization fitness function of the LQG control method in Step 3 is as follows: Step 41: Determine the weight coefficient ratio of vehicle body acceleration, tire dynamic load, and suspension dynamic deflection through the analytic hierarchy process to clarify the optimization priority of each performance parameter. When not exceeding the threshold, set its weight coefficient ratio to 5:1:0; when exceeding the threshold, set its weight coefficient ratio to 7:1;1 Step 42: Design the penalty coefficient Ξ = 10 to improve the optimization priority when the index exceeds the threshold; Step 43: According to 41, 42 and Step 2, construct the optimization fitness function for each performance index: Construct the body acceleration fitness function R1 according to the subjective feeling of the human body Construct the tire dynamic load fitness function R2 Construct the suspension dynamic deflection fitness function R3 Then, the total fitness function is R = R1 + R2 + R3.

4. The method according to claim 1, wherein The method for constructing the optimal feedback matrix K of the LQG control in Step 5 is as follows: Solve through the LQR function in MATLAB to calculate the optimal feedback matrix K.

Images