Smith pre-estimation compensation variable universe fuzzy PID electro-hydraulic suspension position control method

Abstract

The invention discloses a Smith pre-estimation compensation variable universe fuzzy PID electro-hydraulic suspension position control method, and belongs to the field of vehicle active suspension control. Aiming at the problems of position tracking lag and vibration reduction performance degradation caused by main time lag, strong nonlinearity and parameter fluctuation of a valve control cylinder servo system, a coupling dynamic model of a quarter vehicle two-degree-of-freedom suspension and an electro-hydraulic servo actuator is constructed, and an equivalent main time lag parameter is measured through multiple simulation experiments. And a Smith pre-estimation link is adopted to carry out predictive compensation on a time delay channel, meanwhile, a control error and an error change rate are used for driving variable universe fuzzy reasoning to carry out online self-adaptive fine adjustment and correction on PID parameters, and servo valve control voltage is output. Random road surface displacement excitation is further introduced to carry out simulation comparison verification, sprung mass displacement and speed and acceleration response are integrated, tracking precision, disturbance suppression capability and robustness are remarkably improved, and vehicle smoothness and riding comfort are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of vehicle electro-hydraulic servo active suspension control, and particularly relates to a control method for realizing position accurate tracking of an electro-hydraulic servo active suspension under random road excitation and external disturbance, in particular to a Smith prediction compensation variable universe fuzzy PID electro-hydraulic suspension position control method. BACKGROUND

[0002] A vehicle suspension system is a key component connecting a vehicle body and wheels, and its performance directly affects vehicle driving safety, steering stability and ride comfort. A traditional passive suspension mainly relies on fixed parameters of spring and damping elements to absorb road excitation, and its control ability is limited, and it is difficult to realize active adjustment according to changes in vehicle speed, load and road conditions. Under the action of random road excitation or impact load, the passive suspension often shows phenomena such as increased body displacement and vibration response, obvious attitude change and long recovery process, and it is difficult to achieve a good balance between safety and comfort, which restricts the improvement of the overall performance of the vehicle.

[0003] To improve the above problems, an electro-hydraulic servo active suspension outputs controllable force or displacement through a valve-controlled hydraulic actuator, realizes active adjustment of the suspension dynamic response, and can improve the adaptability of the vehicle to complex road conditions to a certain extent. However, the electro-hydraulic servo system usually has nonlinear and strong coupling characteristics, and the system dynamics is jointly affected by factors such as compressibility of hydraulic oil, valve port flow characteristics, leakage and friction, resulting in uncertainty of model parameters with changes in temperature, pressure and load working conditions, further increasing the control difficulty. At the same time, the main time delay is inevitably introduced in the control channel by hydraulic transmission, valve core dynamics, signal acquisition and calculation links, which makes the control input lag behind the output, reduces the closed-loop phase margin and response speed, and easily causes tracking accuracy to decrease and stability margin to be insufficient, which is more prominent when the random road excitation acts.

[0004] However, in the prior art, PID, fuzzy PID and part of modern control methods are more applied in engineering, which has the characteristics of low implementation cost and relatively clear structure, but when the main time delay, random road excitation and parameter uncertainty exist at the same time, it is often difficult to balance rapidity, steady-state accuracy and robustness. On the one hand, the PID parameters are fixed, which is difficult to adapt to the change of dynamic characteristics under different working conditions, and is easy to increase overshoot, slow convergence or increase steady-state error; on the other hand, although fuzzy PID has certain online adjustment ability, if the domain range is fixed or the adjustment mechanism does not adaptively change with the error state, it is easy to appear insufficient adjustment or excessive adjustment, thereby causing limited vibration suppression ability or control quality decline. Therefore, it is necessary to propose a control method that can consider the influence of the main time delay of the electro-hydraulic servo active suspension and the nonlinear characteristics of the system, predict and compensate the main time delay to improve the closed-loop stability and response speed, and realize online updating of PID parameters combined with variable domain fuzzy self-tuning mechanism, so as to improve the position tracking accuracy and enhance the vibration suppression ability of the vehicle body under the conditions of random road and external disturbance, and realize better comprehensive performance of safety, stability and comfort. SUMMARY

[0005] To solve the problems in the background art that the electro-hydraulic servo active suspension is easily affected by the main time delay under the conditions of random road excitation and external disturbance, the position tracking accuracy is insufficient due to the system nonlinear and parameter uncertainty, the closed-loop stability margin is reduced, and the vehicle body vibration suppression ability is limited, the present application proposes a Smith prediction compensation variable domain fuzzy PID electro-hydraulic suspension position control method. Based on the establishment of the dynamic model of the two-degree-of-freedom active suspension of the quarter vehicle, the valve-controlled cylinder electro-hydraulic servo actuator model is constructed, which is driven by the servo valve control input and the output of the hydraulic cylinder piston displacement, and the actuator model is coupled with the suspension dynamic model to form the controlled object containing the main time delay parameter; further, the Smith prediction compensation structure is used to predict and compensate the main time delay, and the PID parameters are online self-tuned through variable domain fuzzy reasoning, so as to improve the position tracking accuracy and closed-loop stability under the conditions of random road and disturbance, reduce the responses of the sprung mass displacement, speed and acceleration, and improve the safety, stability and ride comfort of the whole vehicle.

[0006] Specifically, the present application provides a Smith prediction compensation variable domain fuzzy PID electro-hydraulic suspension position control method, characterized in that it comprises the following steps:

[0007] Step 1: Obtain the vehicle parameters and running state data, and establish the dynamic model of the two-degree-of-freedom electro-hydraulic active suspension of the quarter vehicle;

[0008] Step two: a valve-controlled cylinder electro-hydraulic servo actuator model is established with the servo valve control voltage as input and the hydraulic cylinder piston displacement as output, the main time delay parameter of the electro-hydraulic servo control channel is determined, the actuator model is coupled with the suspension dynamics model, and a controlled object of the electro-hydraulic active suspension containing time delay is constructed;

[0009] Step three: on the basis of the controlled object containing time delay, a Smith prediction compensation-variable universe fuzzy PID parameter self-tuning compound controller is constructed; wherein the Smith prediction compensation structure is used to predict and compensate the main time delay, and the PID control parameter is self-tuned online through variable universe fuzzy reasoning to generate the servo valve control voltage to realize the closed-loop control of the electro-hydraulic active suspension;

[0010] Step four: a random road excitation model is established and a random road excitation signal is generated, the random road excitation signal is taken as the road input of the dynamics model, and a simulation excitation condition for performance verification is constructed;

[0011] Step five: under the random road excitation and external disturbance working condition, the online parameter tuning rules of the controller parameter, the fuzzy rule and the universe expansion factor of the compound controller are finely adjusted adaptively, and the simulation results are compared and analyzed to verify the control effect and robustness of the control method.

[0012] Further improvement of the technical scheme of the application is that the step one comprises the following contents.

[0013] S1. Obtain the state information of the vehicle hydraulic active suspension system, the state information comprising state input data and state model data for describing the dynamic characteristics of the system;

[0014] S2. Based on the state model data obtained in step S1, a two-degree-of-freedom dynamics model of the electro-hydraulic servo active suspension is established, which is specifically as follows.

[0015] S21. The dynamics model of the active suspension is established according to Newton's second law:

[0016]

[0017] In the above formula, F t The expression is as follows:

[0018]

[0019] In the suspension dynamics model, m s represents the total mass of the sprung mass and the passenger mass; m u represents the unsprung mass, i.e. the total mass of the tire assembly; B p represents the damping coefficient of the suspension system; K sIndicates the stiffness coefficient of the suspension system; x ms Indicates the vertical displacement of the sprung mass; x mu Indicates the vertical displacement of the unsprung mass; x p F represents the output displacement of the electro-hydraulic actuator system. t K1 and K2 represent the elastic force of the tire; K1 and K2 represent the elastic force coefficients under various working conditions, and neither of them is zero. This represents the random road surface excitation input.

[0020] S22. Abstract the dynamic model into a mathematical model of the suspension;

[0021] First, the dynamic model is written in the form of a state-space expression:

[0022] Establish the spatial state expression for the vehicle's active suspension, defining the state variables as follows:

[0023] , , ,

[0024] The dynamic equation above can be rewritten as:

[0025]

[0026] A further improvement to the technical solution of the present invention is that step two includes the following:

[0027] S3. Based on the vehicle structural parameters and operating status data obtained in step S1, and the quarter-vehicle two-degree-of-freedom electro-hydraulic active suspension dynamic model established in step S2, a nonlinear dynamic model of the valve-controlled cylinder electro-hydraulic servo actuator is established with servo valve control voltage as input and hydraulic cylinder piston displacement as output. Using piston displacement, piston speed, and load pressure as state variables, the valve orifice throttling load flow equation, load flow continuity equation, piston load system dynamic equation, and the servo valve voltage and valve core displacement driving relationship are established sequentially, and the main time delay parameter of the electro-hydraulic servo control channel is identified. The actuator model is coupled with the suspension dynamic model to construct an electro-hydraulic active suspension controlled object containing the main time delay element, providing an object model basis for the subsequent design of the Smith prediction compensation and variable universe fuzzy self-tuning PID composite controller, as detailed below:

[0028] S31. Construct the actuator load flow equation generated by throttling at the servo valve orifice:

[0029]

[0030] Among them, C dLet ω be the flow coefficient at the throttling orifice, and ω be the area gradient of the throttling window of the slide valve. n = A1 / A2, where A1 is the piston area of ​​the rodless chamber, A2 is the piston area of ​​the rod chamber, and x... vr Where ρ is the valve core opening and P is the liquid density. s For oil supply pressure, P L This refers to the load pressure.

[0031] S32. Construct the linearized model equation for the valve orifice load flow:

[0032]

[0033] Wherein, the flow gain K in the linearized model q With pressure gain K c The load flow equation is obtained by performing small-signal linearization at the operating point, which satisfies the following conditions: , When the piston rod extends, the K q With K c The specific expressions are as follows:

[0034]

[0035] Similarly, when the system piston rod retracts, the flow rate and pressure gain are respectively K q =K qh K c =K ch .

[0036] S33. Establish the continuity constraints and dynamic relationships of the valve-controlled asymmetric hydraulic cylinder electro-hydraulic servo actuator under the piston rod extension condition, and construct the actuator load flow continuity equation and the piston load system dynamic equation respectively:

[0037]

[0038] Among them, v t Let βe be the effective volume of the hydraulic system, and x be the equivalent bulk modulus of the hydraulic fluid. p Let m be the displacement of the piston extension in the system, and B be the equivalent mass. p Let F be the equivalent viscous damping coefficient, and K be the equivalent stiffness coefficient. External disturbance force F L Caused by the state variables of the suspension sprung and unsprung loads, and can be expressed as:

[0039]

[0040] Where A is the effective pressure-bearing area of ​​the hydraulic cylinder, in meters. u For unsprung mass, C r K is the equivalent coefficient of suspension damping. s K is the equivalent stiffness coefficient of the suspension spring.cc This represents the tire's equivalent stiffness coefficient, and x1 to x4 are suspension state variables.

[0041] S34. Construct the driving relationship between the servo valve spool displacement and the control input voltage:

[0042] Assuming the servo valve spool displacement is related to the control input voltage u r The linear proportional relationship between them can be expressed as:

[0043]

[0044] Among them, K v This is the voltage-displacement gain coefficient of the servo valve, which is used to convert the electrical signal u r This is mapped to the adjustment amount of hydraulic flow.

[0045] Based on the above valve core displacement-voltage drive relationship, the aforementioned linearized load flow model, load flow continuity equation, and piston dynamics equation, and by combining and simplifying these equations and eliminating the load flow and load pressure variables, the dynamic equation for the extension / retraction displacement of the hydraulic cylinder piston rod can be obtained.

[0046] S35. Construct the system dynamic equations under the piston rod extension condition:

[0047]

[0048] in, .

[0049] S36. Construct the system dynamic equations under the piston rod extension condition:

[0050]

[0051] in, .

[0052] S37. Construct the state differential equations of the hydraulic system:

[0053] Choose piston displacement, piston velocity, and piston acceleration as state variables, and let:

[0054] , ,

[0055] The state differential equations of the hydraulic system can be derived from the above system dynamic equations as follows:

[0056]

[0057] A further improvement to the technical solution of the present invention is that step three includes the following:

[0058] S4. Based on the vehicle structural parameters and operating status data obtained in step S1, and the controlled object model of the valve-controlled cylinder electro-hydraulic servo actuator with main time delay and the quarter-vehicle two-degree-of-freedom active suspension coupled in step S3, in order to weaken the adverse effects of the main time delay on closed-loop stability and position tracking accuracy and to achieve online self-tuning of control parameters, a Smith prediction compensation structure is constructed to predict and compensate for the main time delay link, forming a predictive output for feedback; the control error and error change rate are constructed with the expected displacement and the predicted output, and an online tuning mechanism for PID parameters with variable domain fuzzy inference is established to dynamically adjust the proportional, integral and derivative parameters and generate the servo valve control input voltage to drive the valve-controlled cylinder electro-hydraulic servo actuator to achieve closed-loop accurate tracking of the hydraulic cylinder piston displacement, and to improve the system tracking performance and robustness under random road excitation and parameter disturbance conditions, as detailed below:

[0059] S41. Construct the Smith prediction compensation structure:

[0060] Based on the coupled controlled object with main time delay obtained in step S3, a Smith prediction and compensation structure is constructed to achieve prediction and compensation for the main time delay element; wherein, the controlled object is represented as consisting of a time-delay-free part With pure delay A series-connected object model is used, and an outer loop controller is set. Simultaneously, an internal model of the Smith predictor is constructed, including a time-delay-free model. and its corresponding time delay model Based on this, the transfer function of the compensation stage is constructed:

[0061]

[0062] in, For the expected input, For error signals, For controller output, This is the system output. Based on the Smith prediction compensation structure, the closed-loop transfer function can be expressed as:

[0063]

[0064] When the internal model is dynamically consistent with the controlled object and the time delay parameters are accurately identified, that is, when the internal model is consistent with the controlled object and the time delay parameters are accurately identified, the following conditions are met. and If the actual master time delay is consistent, then the closed-loop transfer function can be equivalent to:

[0065]

[0066] At this point, the closed-loop characteristic equation can be further expressed in an equivalent form as:

[0067]

[0068] This results in the closed-loop poles being jointly determined by the time-delay-free object and the outer-loop controller, and the main time-delay term no longer entering the closed-loop characteristic polynomial; the system output only exhibits a pure delay element e. -τs Therefore, the Smith prediction compensation structure can weaken the adverse effects of the master time delay on system stability and tracking performance, and realize prediction compensation for elements containing master time delay.

[0069] S42. Constructing a universe-of-discourse scaling mechanism for variable universe-of-discourse fuzzy self-tuning PID:

[0070] To avoid insufficient resolution in the small error region due to a fixed initial universe of discourse, which could lead to a decrease in steady-state accuracy, a dynamic universe of discourse adjustment mechanism is introduced into the variable universe of discourse fuzzy self-tuning PID: based on the predicted output formed by the Smith prediction compensation structure, the control error is constructed. and error change rate and will , As input for fuzzy inference; let the input variable be... , The initial universes of discourse are [−E, E] and [−E], respectively. C E C The initial universe of discourse for the output variable u is [−U, U]. Without altering the shape of the membership function and the fuzzy rules, a universe of discourse scaling factor is introduced to dynamically scale the input / output universes, resulting in the scaled universes expressed as follows:

[0071]

[0072] in, , For the input universe of discourse scaling factor, To output the universe of discourse scaling factor; when or When the domain is large, the dynamic response capability is improved by expanding the domain of discourse. and When the size is small, the domain of discourse is contracted to improve the accuracy of steady-state control.

[0073] The universe of discourse scaling factor is driven by the error magnitude and error variation trend, and the error threshold is assumed to satisfy... ,in , Assume the error rate of change threshold satisfies ,in , And set upper and lower limits for the scaling factor to meet the requirements. ,in The value ranges from 0.3 to 0.6. The value ranges from 1.0 to 2.0; therefore, the input scaling factor can be determined according to the following piecewise linear rule:

[0074]

[0075] Through the aforementioned dynamic scaling rules, the input universe of discourse automatically expands when the system is in a large error phase to enhance the adjustment amplitude and response speed; as the error gradually decreases and approaches zero, the input universe of discourse automatically shrinks to improve the control resolution and steady-state accuracy in the small error region, thereby achieving a balance between dynamic and steady-state performance. Furthermore, the input linguistic variable set is preferably divided into {NB, NM, NS, ZO, PS, PM, PB}, and the membership function is preferably a triangular membership function to ensure the continuity and feasibility of fuzzy inference under dynamic scaling conditions.

[0076] S43. Constructing the rules for determining the universe of discourse scaling factor:

[0077] To achieve online, refined, and adaptive regulation of fuzzy self-tuning PID controllers with variable universe of discourse, a function-based method for constructing the universe of discourse scaling factor is adopted. The selected function must satisfy the following five conditions: consistency, monotonicity, normality, duality, and zero-avoidance. The function satisfying these five conditions can be defined as follows:

[0078]

[0079] Where x is the input variable; λ is the function coefficient, usually λ∈(0,1). As λ increases, α(x) decreases, but the change in α(x) is more drastic, the universe of discourse compression is obvious, and the system response is fast; K is the exponential coefficient and K>0, the larger K is, the larger α(x) is. K is the coefficient of the integral term; n is the number of input variables, here taken as 2; p i It is the i-th element in the constant vector; e i β is the i-th element in the input deviation vector; β(0) is the initial value of the output variable's universe of discourse scaling factor. Using the above functional form, the universe of discourse can be expanded to enhance dynamic response when the error is large, and contracted to improve steady-state control accuracy when the error approaches zero.

[0080] Based on the above principles and through multiple experiments, the input universe scaling can be set as follows:

[0081]

[0082] Through analysis , and After assessing the system's control effect, the universe of discourse scaling factor for the output variable was determined. The results show that... and monotony and The monotonicity is consistent. Conversely, monotony and The monotonicity is opposite. Therefore, after many experiments, the following scaling factor was chosen and can be expressed as:

[0083]

[0084] in: , , Position control , , The output universe of discourse scaling factor.

[0085] S43. Constructing the process of fuzzy inference, defuzzification, and control law output:

[0086] Based on the fuzzification of input variables and the establishment of a fuzzy rule base, the Mamdani-type inference method is used to represent the fuzzy conditional statement "if A then B" as a fuzzy relation R from the input domain to the output domain. When the input is A... * When, output fuzzy set B * Obtained by fuzzy synthesis:

[0087]

[0088] in," "This is a fuzzy composition operator. Therefore, the fuzzy Ra→ε relation Mamdani is based on the definition:

[0089]

[0090] The membership function of the output fuzzy set is obtained by using the supremum-minimum composition rule:

[0091]

[0092] To obtain a definite quantity that can be used for online parameter tuning, the output fuzzy set is defuzzified; preferably, the centroid method is used to obtain a clear output u. c ,satisfy:

[0093]

[0094] Alternatively, a weighted average can be used in the discretization implementation:

[0095]

[0096] Based on the defuzzification results, the PID parameter correction ΔK is obtained respectively. p ΔK i ΔKd And it is superimposed on the PID initial parameters to achieve online updates:

[0097]

[0098] in, , , All are initial setpoints for PID control; based on online updates. , , Construct control laws and output control quantities It can be represented as:

[0099]

[0100] and the As the drive input of the valve-controlled cylinder electro-hydraulic servo actuator, the Smith prediction compensation and variable universe fuzzy self-tuning PID work together to reduce phase lag and stability degradation caused by the master time delay. Under random road excitation disturbance and parameter uncertainty, it can achieve high-precision tracking of the desired displacement, effectively suppress vehicle body mass vibration, and improve the system's anti-disturbance and robustness.

[0101] A further improvement to the technical solution of the present invention is that step four includes the following:

[0102] S5. Based on the quarter-vehicle two-degree-of-freedom electro-hydraulic servo active suspension dynamic model established in step S2 and the coupled controlled object model with master time delay obtained in step S3, in order to make the control method verification conditions close to the actual vehicle driving road surface and improve the credibility of simulation demonstration, a random road surface input model is constructed and random road surface displacement input is generated. , will the The road surface displacement input is used as the model of the coupled controlled object, thereby forming the random road surface excitation conditions for performance verification; the establishment and signal generation process of the random road surface input model is as follows:

[0103] S51. Constructing a spatial power spectral density model for road surface irregularities:

[0104] The spatial power spectral density function of road surface roughness is established with spatial frequency n as the independent variable, satisfying the power function form:

[0105]

[0106] Where: n is the spatial frequency; n0 is the reference spatial frequency; This refers to the road surface roughness coefficient. For frequency exponents, the preferred choice is... .

[0107] S52. The definition of the spatial power spectral density of a road surface is:

[0108]

[0109] in: For frequency band The actual power of the included road surface power spectral density.

[0110] S53. Construct the equivalent mapping from the spatial domain to the time domain:

[0111] Let the longitudinal speed of the vehicle be... And establish the correspondence between spatial frequency and temporal frequency as follows:

[0112]

[0113] When the vehicle is at speed During driving, the components of the road surface vertical displacement spectrum contained in the time frequency band and the spatial frequency band are the same, i.e., the frequency band... and The power of all Therefore, the time power spectral density of the road surface displacement is obtained as follows:

[0114]

[0115] In frequency index Under these conditions, the above time power spectral density can be further expressed as:

[0116]

[0117] Therefore, the time power spectral density of the vertical velocity of the road surface can be derived as follows:

[0118]

[0119] As can be seen from the above, the time power spectral density of the road surface vertical velocity is a constant value, possessing the same characteristics as the power spectrum of white noise. Therefore, the white noise shaping method can be used to generate random road surface displacement input.

[0120] S54. Construct a white noise shaping filter and generate random road surface displacement input:

[0121] Take a Gaussian white noise signal with a unit intensity of 1 For input, road surface displacement For the output, let the frequency response function of the shaping filter be... Then the power spectral density of the road surface displacement satisfies:

[0122]

[0123] in: Let the power spectral density of the white noise signal satisfy the following condition:

[0124]

[0125] And take: .

[0126] angular frequency express At that time, the power spectral density of the road surface displacement can be written as:

[0127]

[0128] Therefore, the frequency response function of the shaping filter is:

[0129]

[0130] Therefore, the transfer function of the shaping filter is obtained as follows:

[0131]

[0132] Therefore, the frequency domain relation satisfies:

[0133]

[0134] Converting the above frequency domain relationship to the time domain yields the following road surface time domain input model:

[0135]

[0136] Since the road surface spectrum is approximately horizontal in the low-frequency range, a lower cutoff frequency is introduced. The time power spectral density of the road surface displacement is updated as follows:

[0137]

[0138] Accordingly, the frequency response function of the shaping filter is updated as follows:

[0139]

[0140] Therefore, the time-domain model of the filtered white noise road surface is:

[0141]

[0142] Furthermore, to avoid introducing deviations by using time frequency for the lower cutoff frequency, the lower cutoff time frequency is... Convert to lower cutoff space frequency The relationship between the two satisfies:

[0143]

[0144] Based on this, the updated random time-domain input model for the road surface is obtained as follows:

[0145]

[0146] in: This refers to the road surface roughness coefficient. The lower cutoff spatial frequency; Reference spatial frequency; The longitudinal speed of the vehicle; This is a white noise signal with an expected value of 0.

[0147] S55. Generate random road conditions and couple them with the input:

[0148] The random road surface displacement excitation signal The road displacement is used as the input for the one-quarter vehicle two-degree-of-freedom electro-hydraulic servo active suspension dynamic model established in step S2, and together with the controlled object model with master time delay coupling obtained in step S3, it constitutes the random road disturbance condition, which is used to verify the closed-loop tracking performance and robustness of the control method under random road excitation conditions.

[0149] S6. A further improvement of the technical solution of the present invention is that step five includes the following:

[0150] A: Under the condition of preset random road surface excitation and vehicle longitudinal speed, external disturbances are superimposed to establish an online adaptive fine adjustment mechanism characterized by the error and error change rate between the expected longitudinal displacement trajectory of the sprung mass and the actual output. Based on the error and error change rate, variable universe of discourse fuzzy inference is driven to perform online self-tuning and updating of the proportional, integral and derivative parameters of the composite controller, and the adjustment strategy of fuzzy rules and universe of discourse scaling factor are simultaneously adaptively corrected. This enables the composite control strategy to maintain closed-loop stability under the condition of master time delay and parameter uncertainty, reduces the adverse effects of time delay and disturbance on position tracking performance, and improves tracking accuracy and anti-disturbance capability.

[0151] B: Under the conditions of random road surface excitation and external disturbance or impact load, a comparative control strategy is constructed and simulated to verify the results. The comparative control strategy includes at least a passive suspension no-control strategy, a PID control strategy, a fuzzy PID control strategy, and the composite control strategy of the present invention. By comparing and analyzing the response of the sprung mass longitudinal displacement, longitudinal velocity, and longitudinal acceleration, and combining dynamic indicators such as overshoot, settling time, and steady-state error, the differences between the strategies are evaluated from the dimensions of driving safety, vehicle longitudinal stability, and ride comfort, so as to verify the superiority of the present invention in terms of vehicle vibration suppression, position tracking accuracy, and robustness.

[0152] The technological advancements achieved by this invention due to the adoption of the above technical solutions are as follows:

[0153] This invention addresses the problems of decreased position tracking accuracy, response lag, and insufficient vibration suppression in electro-hydraulic servo active suspension systems under conditions of principal time delay, strong system nonlinearity, and parameter uncertainty. Based on a coupled model of a valve-controlled cylinder electro-hydraulic servo actuator and a quarter-vehicle two-degree-of-freedom active suspension, and identifying the principal time delay parameters, a composite control strategy combining Smith predictor compensation and variable universe of discourse fuzzy self-tuning PID is constructed. The Smith predictor performs predictive compensation for the principal time delay, and the variable universe of discourse fuzzy inference drives online tuning of the PID parameters and universe of discourse scaling factor based on the error and error change rate. This achieves high-precision closed-loop tracking of the hydraulic cylinder piston displacement and robust suppression of disturbances, thereby maintaining system stability and improving control quality under complex operating conditions.

[0154] Simulation results show that, under random road surface excitation and external disturbance conditions, compared with passive suspension without control strategy, PID control strategy and fuzzy PID control strategy, the composite control method of the present invention can significantly reduce the longitudinal displacement tracking error of the sprung mass and suppress overshoot and vibration, improve the longitudinal velocity and longitudinal acceleration response characteristics of the sprung mass, and comprehensively improve driving safety, vehicle longitudinal stability and ride comfort. At the same time, it shows stronger robustness and engineering applicability under the conditions of parameter fluctuation and time delay. Attached Figure Description

[0155] Figure 1 This is a flowchart of the present invention;

[0156] Figure 2 This is a simplified structural diagram of the two-degree-of-freedom active suspension electro-hydraulic servo system model of the present invention;

[0157] Figure 3 This is a system architecture diagram of the Smith predictor of the present invention;

[0158] Figure 4 This is a vehicle AHS architecture diagram of the variable universe fuzzy PID composite position control strategy based on Smith predictor according to the present invention;

[0159] Figure 5 This is a schematic diagram of the variable universe dynamic adjustment mechanism and the input-output universe scaling relationship of the present invention;

[0160] Figure 6 This is a schematic diagram of random road surface displacement excitation generation based on random road surface roughness power spectral density modeling according to the present invention;

[0161] Figure 7 This is a schematic diagram comparing the longitudinal displacement response of the sprung mass obtained by using the SVUFMS-PID composite control strategy of the present invention in an embodiment of the present invention and comparing it with other control strategies.

[0162] Figure 8This is a schematic diagram comparing the longitudinal velocity response of the sprung mass obtained by using the SVUFMS-PID composite control strategy of the present invention in an embodiment of the present invention and comparing it with other control strategies;

[0163] Figure 9 This is a schematic diagram comparing the longitudinal acceleration response of the sprung mass using the SVUFMS-PID composite control strategy of this invention and other control strategies, according to an embodiment of the present invention. Detailed Implementation

[0164] Exemplary embodiments, features, and aspects of the present invention will now be described in detail with reference to the accompanying drawings. Although various parameter values ​​and aspects of the embodiments are shown, the simulation process need not be performed with exactly the same parameters and aspects unless specifically indicated.

[0165] The following will combine Figures 1-9 The present invention will be further described in detail with reference to the embodiments:

[0166] The present invention relates to a Smith-predicted compensation variable universe of discourse fuzzy PID electro-hydraulic suspension position control method, such as... Figure 1 As shown, it includes:

[0167] Step 1: Obtain vehicle parameters and operating status, and establish a one-quarter vehicle two-DOF active suspension dynamics model;

[0168] S1. Obtain the state information of the vehicle's hydraulic active suspension system, the state information including state input data and state model data used to describe the dynamic characteristics of the system;

[0169] S2. Based on the state model data obtained in step S1, establish a two-degree-of-freedom dynamic model of the electro-hydraulic servo active suspension, as shown below:

[0170] like Figure 2 The figure shown is a simplified structural diagram of a two-degree-of-freedom active suspension electro-hydraulic servo system model. In the suspension dynamics model, m s The sprung mass is the sum of the mass of the car body frame and the passenger; m u This indicates the unsprung mass, which is the mass of the tire assembly; B p K represents the damping coefficient of the suspension system. s Indicates the stiffness coefficient of the suspension system; x ms Indicates the vertical displacement of the sprung mass; x mu Indicates the vertical displacement of the unsprung mass; x p F represents the output displacement of the electro-hydraulic actuator system. t K1 and K2 represent the elastic force of the tire; K1 and K2 represent the elastic force coefficients under various working conditions, and neither of them is zero. This represents the random road surface excitation input.

[0171] S21: Establish a dynamic model of the active suspension based on Newton's second law:

[0172]

[0173] In the above formula, F t The expression is as follows:

[0174]

[0175] S22. Abstract the dynamic model into a mathematical model of the suspension;

[0176] make , , , ,

[0177] This represents the first state variable. This represents the first state variable. This represents the first state variable. This represents the first state variable.

[0178] The dynamic equation above can be rewritten as:

[0179]

[0180] Step 2: Establish a nonlinear model of the valve-controlled cylinder electro-hydraulic servo actuator with the servo valve control input voltage as input and the hydraulic cylinder piston displacement as output, and couple it with the quarter-vehicle two-degree-of-freedom active suspension dynamics model to construct a controlled object model containing the master time delay parameters;

[0181] S3. Based on the vehicle structural parameters and operating status data obtained in step S1 and the quarter-vehicle two-degree-of-freedom electro-hydraulic active suspension dynamic model established in step S2, a nonlinear model of a valve-controlled cylinder electro-hydraulic servo actuator is established with the servo valve control input voltage as input and the hydraulic cylinder piston displacement as output. This model is then coupled with the suspension dynamic model. The main time delay parameters are identified, and a coupled controlled object model of the electro-hydraulic active suspension containing the main time delay element is constructed. This provides the object model basis for the subsequent design of the Smith prediction compensation and variable universe fuzzy self-tuning PID composite controller, as detailed below:

[0182] S31. Construct the actuator load flow equation generated by throttling at the servo valve orifice:

[0183]

[0184] Among them, C dLet ω be the flow coefficient at the throttling orifice, and ω be the area gradient of the throttling window of the slide valve. n = A1 / A2, where A1 is the piston area of ​​the rodless chamber, A2 is the piston area of ​​the rod chamber, and x... vr Where ρ is the valve core opening and P is the liquid density. s For the oil supply pressure, P L This refers to the load pressure.

[0185] S32. Construct the linearized model equation for the valve orifice load flow:

[0186]

[0187] Wherein, the flow gain K in the linearized model q With pressure gain K c The load flow equation is obtained by performing small-signal linearization at the operating point, which satisfies the following conditions: , When the piston rod extends, the K q With K c The specific expressions are as follows:

[0188]

[0189] Similarly, when the system piston rod retracts, the flow rate and pressure gain are respectively K q =K qh K c =K ch .

[0190] S33. Establish the continuity constraints and dynamic relationships of the valve-controlled asymmetric hydraulic cylinder electro-hydraulic servo actuator under the piston rod extension condition, and construct the actuator load flow continuity equation and the piston load system dynamic equation respectively:

[0191]

[0192] Among them, v t Let βe be the effective volume of the hydraulic system, and x be the equivalent bulk modulus of the hydraulic fluid. p Let m be the displacement of the piston extension in the system, and B be the equivalent mass. p Let F be the equivalent viscous damping coefficient, and K be the equivalent stiffness coefficient. External disturbance force F L Caused by the state variables of the suspension sprung and unsprung loads, and can be expressed as:

[0193]

[0194] Where A is the effective pressure-bearing area of ​​the hydraulic cylinder, in meters. u For unsprung mass, C r K is the equivalent coefficient of suspension damping. s K is the equivalent stiffness coefficient of the suspension spring.cc This represents the tire's equivalent stiffness coefficient, and x1 to x4 are suspension state variables.

[0195] S34. Construct the driving relationship between the servo valve spool displacement and the control input voltage:

[0196] Assuming the servo valve spool displacement is related to the control input voltage u r The linear proportional relationship between them can be expressed as:

[0197]

[0198] Among them, K v This is the voltage-displacement gain coefficient of the servo valve, which is used to convert the electrical signal u r This is mapped to the adjustment amount of hydraulic flow.

[0199] Based on the above valve core displacement-voltage drive relationship, the aforementioned linearized load flow model, load flow continuity equation, and piston dynamics equation, and by combining and simplifying these equations and eliminating the load flow and load pressure variables, the dynamic equation for the extension / retraction displacement of the hydraulic cylinder piston rod can be obtained.

[0200] S35. Construct the system dynamic equations under the piston rod extension condition:

[0201]

[0202] in, .

[0203] S36. Construct the system dynamic equations under the piston rod extension condition:

[0204]

[0205] in,

[0206] S37. Construct the state differential equations of the hydraulic system:

[0207] Choose piston displacement, piston velocity, and piston acceleration as state variables, and let:

[0208] , ,

[0209] The state differential equations of the hydraulic system can be derived from the above system dynamic equations as follows:

[0210]

[0211] Step 3: Based on the coupled controlled object model, construct a Smith prediction compensation and variable universe fuzzy self-tuning PID composite control strategy to provide a controller structure foundation for subsequent closed-loop position tracking and robust vibration suppression control.

[0212] S4. Based on the vehicle structural parameters and operating status data obtained in step S1, combined with the quarter-vehicle two-degree-of-freedom electro-hydraulic servo active suspension dynamic model established in step S2 and the controlled object model with master time delay coupling obtained in step S3, in order to weaken the adverse effects of master time delay on closed-loop stability and position tracking accuracy and improve robustness under random road surface and parameter uncertainty conditions, a Smith prediction compensation link is constructed to predict and compensate the master time delay channel. A variable universe of discourse fuzzy self-tuning PID online tuning mechanism is also constructed. The input and output universe of discourse are dynamically expanded based on the control error and error change rate, and the PID parameter correction is output. This achieves online updating of proportional, integral, and derivative parameters, generating the servo valve control input voltage as the actuator drive quantity, thereby realizing high-precision closed-loop tracking of hydraulic cylinder piston displacement and vehicle body vibration suppression. Specifically, as follows:

[0213] S41. Construct the Smith prediction compensation structure:

[0214] Based on the controlled object model with master time delay coupling obtained in step S3, a Smith prediction and compensation structure is constructed to achieve prediction and compensation of the master time delay element. Its system architecture is as follows: Figure 3 As shown. The controlled object is equivalently represented as the time-delay-free part. With pure delay It is connected in series and an outer loop controller is set. Simultaneously, an internal model of the Smith predictor is constructed, including a time-lag-free model. and its corresponding time-delay model Based on this, the transfer function of the compensation stage is constructed as follows:

[0215]

[0216] in, For the expected input, For error signals, For controller output, This is the system output. Based on the Smith prediction compensation structure, the closed-loop transfer function can be expressed as:

[0217]

[0218] When the internal model is dynamically consistent with the controlled object and the time delay parameters are accurately identified, that is, when the internal model is consistent with the controlled object and the time delay parameters are accurately identified, the following conditions are met. and If the actual master time delay is consistent, then the closed-loop transfer function can be equivalent to:

[0219]

[0220] At this point, the closed-loop characteristic equation can be further expressed in an equivalent form as:

[0221]

[0222] This results in the closed-loop poles being jointly determined by the time-delay-free object and the outer-loop controller, and the main time-delay term no longer entering the closed-loop characteristic polynomial; the system output only exhibits a pure delay element e. -τs Therefore, the Smith prediction compensation structure can weaken the adverse effects of the master time delay on system stability and tracking performance, and realize prediction compensation for elements containing master time delay.

[0223] S42. Construct a universe-scaling mechanism for variable universe-of-discourse fuzzy self-tuning PID based on Smith prediction compensation:

[0224] like Figure 4 The diagram shown illustrates the vehicle AHS architecture of the variable universe of discourse fuzzy PID composite position control strategy based on the Smith predictor according to the present invention. In this composite position control structure, to overcome the decrease in steady-state accuracy caused by insufficient resolution of the fixed initial universe of discourse in the small error region, a dynamic universe of discourse adjustment mechanism is set in the variable universe of discourse fuzzy self-tuning PID. The predictive output based on the Smith predictor compensation structure constructs the control error. With error change rate And it serves as the input for fuzzy inference. Let... and The initial domains are respectively and And the initial universe of discourse for the output variable u is Without altering the membership function and rule base, a universe scaling factor is introduced to dynamically scale the input and output universes, resulting in the scaled universes represented as follows:

[0225]

[0226] in, , For the input universe of discourse scaling factor, To output the universe of discourse scaling factor; when or When the domain is large, the dynamic response capability is improved by expanding the domain of discourse. and When the size is small, the domain of discourse is contracted to improve the accuracy of steady-state control.

[0227] The universe of discourse scaling factor is driven by the error magnitude and error variation trend, and the error threshold is assumed to satisfy... ,in , Assume the error rate of change threshold satisfies ,in , And set upper and lower limits for the scaling factor to meet the requirements. ,in The value ranges from 0.3 to 0.6. The value ranges from 1.0 to 2.0; therefore, the input scaling factor can be determined according to the following piecewise linear rule:

[0228]

[0229] like Figure 5 The diagram illustrates the dynamic adjustment mechanism of the variable universe of discourse and the scaling relationship between the input and output universes of discourse in this invention. Through the scaling rules shown in the dynamic diagram, when the system is in a large error phase, the input universe of discourse automatically expands to enhance the adjustment amplitude and response speed; when the error gradually decreases and approaches zero, the input universe of discourse automatically shrinks to improve the control resolution and steady-state accuracy in the small error region, thereby achieving a balance between dynamic and steady-state performance. Furthermore, the set of input linguistic variables is preferably divided into {NB, NM, NS, ZO, PS, PM, PB}, and the membership function is preferably a triangular membership function to ensure the continuity and feasibility of fuzzy inference under dynamic scaling conditions.

[0230] S43. Constructing the rules for determining the universe of discourse scaling factor:

[0231] To achieve online, refined, and adaptive regulation of fuzzy self-tuning PID controllers with variable universe of discourse, a function-based method for constructing the universe of discourse scaling factor is adopted. The selected function must satisfy the following five conditions: consistency, monotonicity, normality, duality, and zero-avoidance. The function satisfying these five conditions can be defined as follows:

[0232]

[0233] Where x is the input variable; λ is the function coefficient, usually λ∈(0,1). As λ increases, α(x) decreases, but the change in α(x) is more drastic, the universe of discourse compression is obvious, and the system response is fast; K is the exponential coefficient and K>0, the larger K is, the larger α(x) is. K is the coefficient of the integral term; n is the number of input variables, here taken as 2; p i It is the i-th element in the constant vector; e i β is the i-th element in the input deviation vector; β(0) is the initial value of the output variable's universe of discourse scaling factor. Using the above functional form, the universe of discourse can be expanded to enhance dynamic response when the error is large, and contracted to improve steady-state control accuracy when the error approaches zero.

[0234] Based on the above principles and through multiple experiments, the input universe scaling can be set as follows:

[0235]

[0236] Through analysis , and After assessing the system's control effect, the universe of discourse scaling factor for the output variable was determined. The results show that... and monotony and The monotonicity is consistent. Conversely, monotony and The monotonicity is opposite. Therefore, after many experiments, the following scaling factor was chosen and can be expressed as:

[0237]

[0238] in: , , Position control , , The output universe of discourse scaling factor.

[0239] S43. Constructing the process of fuzzy inference, defuzzification, and control law output:

[0240] Based on the fuzzification of input variables and the establishment of a fuzzy rule base, the Mamdani-type inference method is used to represent the fuzzy conditional statement "if A then B" as a fuzzy relation R from the input domain to the output domain. When the input is A... * When, output fuzzy set B * Obtained by fuzzy synthesis:

[0241]

[0242] in," "This is a fuzzy composition operator. Therefore, the fuzzy Ra→ε relation Mamdani is based on the definition:

[0243]

[0244] The membership function of the output fuzzy set is obtained by using the supremum-minimum composition rule:

[0245]

[0246] To obtain a definite quantity that can be used for online parameter tuning, the output fuzzy set is defuzzified; preferably, the centroid method is used to obtain a clear output u. c ,satisfy:

[0247]

[0248] Alternatively, a weighted average can be used in the discretization implementation:

[0249]

[0250] Based on the defuzzification results, the PID parameter correction ΔK is obtained respectively. p ΔK i ΔK d And it is superimposed on the PID initial parameters to achieve online updates:

[0251]

[0252] in, , , All are initial setpoints for PID control; based on online updates. , , Construct control laws and output control quantities It can be represented as:

[0253]

[0254] and the As the drive input of the valve-controlled cylinder electro-hydraulic servo actuator, the Smith prediction compensation and variable universe fuzzy self-tuning PID work together to reduce phase lag and stability degradation caused by the master time delay. Under random road excitation disturbance and parameter uncertainty, it can achieve high-precision tracking of the desired displacement, effectively suppress vehicle body mass vibration, and improve the system's anti-disturbance and robustness.

[0255] Step 4: Construct a random road surface input model and generate a random road surface displacement excitation signal. Use the random road surface displacement excitation signal as the road surface displacement input of the controlled object model with master time delay coupling, thereby forming a simulation excitation condition for performance verification.

[0256] S5. Based on the quarter-vehicle two-degree-of-freedom electro-hydraulic servo active suspension dynamic model established in step S2 and the controlled object model with master time delay coupling obtained in step S3, in order to make the control method verification conditions close to the actual vehicle driving road surface and improve the credibility of simulation demonstration, a random road surface input model is constructed and random road surface displacement input is generated. and the The process of establishing the random road surface input model and generating signals, which serves as the road surface displacement input for the coupled controlled object model, is as follows:

[0257] S51. Constructing a spatial power spectral density model for road surface irregularities:

[0258] The spatial power spectral density function of road surface roughness is established with spatial frequency n as the independent variable, satisfying the power function form:

[0259]

[0260] Where: n is the spatial frequency; n0 is the reference spatial frequency; This refers to the road surface roughness coefficient. For frequency exponents, the preferred choice is... .

[0261] S52. The definition of the spatial power spectral density of a road surface is:

[0262]

[0263] in: For frequency band The actual power of the included road surface power spectral density.

[0264] S53. Construct the equivalent mapping from the spatial domain to the time domain:

[0265] Let the longitudinal speed of the vehicle be... And establish the correspondence between spatial frequency and temporal frequency as follows:

[0266]

[0267] When the vehicle is at speed During driving, the components of the road surface vertical displacement spectrum contained in the time frequency band and the spatial frequency band are the same, i.e., the frequency band... and The power of all Therefore, the time power spectral density of the road surface displacement is obtained as follows:

[0268]

[0269] In frequency index Under these conditions, the above time power spectral density can be further expressed as:

[0270]

[0271] Therefore, the time power spectral density of the vertical velocity of the road surface can be derived as follows:

[0272]

[0273] As can be seen from the above, the time power spectral density of the road surface vertical velocity is a constant value, possessing the same characteristics as the power spectrum of white noise. Therefore, the white noise shaping method can be used to generate random road surface displacement input.

[0274] S54. Construct a white noise shaping filter and generate random road surface displacement input:

[0275] Take a Gaussian white noise signal with a unit intensity of 1 For input, road surface displacement For the output, let the frequency response function of the shaping filter be... Then the power spectral density of the road surface displacement satisfies:

[0276]

[0277] in: Let the power spectral density of the white noise signal satisfy the following condition:

[0278]

[0279] And take:

[0280] angular frequency express At that time, the power spectral density of the road surface displacement can be written as:

[0281]

[0282] Therefore, the frequency response function of the shaping filter is:

[0283]

[0284] Therefore, the transfer function of the shaping filter is obtained as follows:

[0285]

[0286] Therefore, the frequency domain relation satisfies:

[0287]

[0288] Converting the above frequency domain relationship to the time domain yields the following road surface time domain input model:

[0289]

[0290] Since the road surface spectrum is approximately horizontal in the low-frequency range, a lower cutoff frequency is introduced. The time power spectral density of the road surface displacement is updated as follows:

[0291]

[0292] Accordingly, the frequency response function of the shaping filter is updated as follows:

[0293]

[0294] Therefore, the time-domain model of the filtered white noise road surface is:

[0295]

[0296] Furthermore, to avoid introducing deviations by using time frequency for the lower cutoff frequency, the lower cutoff time frequency is... Convert to lower cutoff space frequency The relationship between the two satisfies:

[0297]

[0298] Based on this, the updated random time-domain input model for the road surface is obtained as follows:

[0299]

[0300] in: This refers to the road surface roughness coefficient. The lower cutoff spatial frequency; Reference spatial frequency; The longitudinal speed of the vehicle; This is a white noise signal with an expected value of 0.

[0301] S55. Generate random road conditions and couple them with the input:

[0302] like Figure 6 The diagram illustrates the generation of random road surface displacement excitation based on the power spectral density of random road surface roughness, and the generation of random road surface displacement excitation signals accordingly. ; will the The road displacement is used as the input for the one-quarter vehicle two-degree-of-freedom electro-hydraulic servo active suspension dynamic model established in step S2, and together with the controlled object model with master time delay coupling obtained in step S3, it constitutes the random road disturbance condition, which is used to verify the closed-loop tracking performance and robustness of the control method under random road excitation conditions.

[0303] Step 5: Under the simulated working conditions formed by the random road surface displacement excitation signal and external disturbance, the controller parameters, fuzzy rules and universe of discourse scaling factor of the composite controller are adjusted online adaptively and finely, and comparative simulation verification is carried out; by comparing and analyzing the longitudinal displacement, longitudinal velocity and longitudinal acceleration response of the sprung mass, the vibration reduction and disturbance rejection robustness of the control strategy is evaluated, and the position tracking compensation and time delay compensation effects are indirectly characterized accordingly.

[0304] S6. Based on the quarter-vehicle two-degree-of-freedom electro-hydraulic servo active suspension dynamic model established in step S2, the controlled object model with master time delay coupling obtained in step S3, and the random road surface displacement excitation signal generated in step S5, comparative simulations are conducted under preset random road surface excitation and vehicle driving conditions with external disturbances superimposed. The longitudinal displacement, longitudinal velocity, and longitudinal acceleration response of the sprung mass are used as evaluation indicators to comprehensively evaluate the vibration reduction and disturbance rejection robustness of the composite control strategy, and to indirectly characterize its position tracking and time delay compensation effects, as detailed below:

[0305] S61. Comparison of controller parameter adjustment and simulation results:

[0306] The values ​​of suspension system parameters and controller parameters;

[0307] Simulation verification

[0308] The parameters of the active suspension system are shown in Table 1.

[0309] Table 1

[0310] To closely approximate the stochastic broadband characteristics of actual road roughness, this embodiment employs stochastic road excitation based on road roughness power spectral density modeling to generate road displacement input. This was used as the road displacement input for the active suspension dynamics model for comparative simulation verification. A schematic diagram of the random road excitation input is shown below. Figure 6 As shown.

[0311] In this embodiment, the initial setpoint parameters for the PID are set. , , The equivalent principal time delay parameters were determined using multiple simulation experiments. As the time delay basis for Smith's estimated compensation, the controller parameter values ​​are as follows:

[0312] , , ,

[0313] Based on the initial given parameters and equivalent master time delay parameters of the SVUFMS-PID controller mentioned above, a Smith prediction compensation and variable universe fuzzy self-tuning PID composite controller was constructed and its parameters were tuned. Comparative simulation verification was carried out under random road excitation and external disturbance conditions. The vibration reduction performance and robustness of the control method of the present invention were comprehensively evaluated from indicators such as the longitudinal displacement response, longitudinal velocity response and longitudinal acceleration response of the sprung mass.

[0314] S62. Adjust the controller and analyze the effect of the SVUFMS-PID control strategy on the vehicle's active suspension system:

[0315] Select random road surface displacement excitation signal The road surface input is:

[0316]

[0317] The following will further illustrate the controller's control effect through graphical comparisons, such as... Figures 7-9 As shown, to verify the vibration reduction compensation effect of the SVUFMS-PID composite control strategy based on Smith prediction compensation and variable universe fuzzy self-tuning PID proposed in this invention under the conditions of random road disturbance and external disturbance superposition, the longitudinal displacement, longitudinal velocity, and longitudinal acceleration of the sprung mass are selected as evaluation indicators. The dynamic response differences of passive suspension, conventional PID control strategy, fuzzy PID control strategy and the SVUFMS-PID composite control strategy of this invention are compared from the dimensions of driving safety, vehicle posture stability and ride comfort.

[0318] like Figure 7 As shown, under random road disturbances, the longitudinal displacement of the sprung mass of the passive suspension varies from -0.076m to 0.058m. After adopting a conventional PID control strategy, the displacement range converges to -0.018m to 0.017m. After adopting a fuzzy PID control strategy, the displacement range further converges to -0.016m to 0.015m. After adopting the SVUFMS-PID composite control strategy of this invention, the displacement range is further compressed to -0.006m to 0.004m. The above analysis results indicate that, under the coupling effect of the main time delay and disturbance, this invention uses Smith prediction compensation to predict and correct the time delay element, and uses variable universe fuzzy self-tuning to correct the PID gain online. This allows the equivalent control input of the actuator to promptly offset the displacement deviation caused by road disturbances, thereby significantly reducing sprung mass displacement fluctuations and improving the effectiveness of position compensation and driving safety.

[0319] like Figure 8As shown, under random road disturbances, the longitudinal velocity variation range of the sprung mass of the passive suspension remains stable within -0.400 m / s to 0.400 m / s. With a conventional PID control strategy, the velocity variation range is -0.089 m / s to 0.081 m / s. With a fuzzy PID control strategy, the velocity variation range is 0.048 m / s to 0.051 m / s. Using the SVUFMS-PID composite control strategy of this invention, the velocity variation range further converges to -0.012 m / s to 0.009 m / s. This demonstrates that this invention can effectively suppress sprung mass velocity fluctuations under the influence of disturbance inputs and parameter fluctuations, reduce the amplitude of velocity response fluctuations, improve dynamic response consistency, and make vehicle body posture changes smoother, thereby improving vehicle longitudinal stability and handling smoothness.

[0320] like Figure 9 As shown, the longitudinal acceleration of the sprung mass is used to characterize the vehicle body vibration intensity and reflect the level of ride comfort. The longitudinal acceleration of the sprung mass of the passive suspension varies from -5.146 m / s² to 4.853 m / s². With a conventional PID control strategy, the acceleration ranges from -2.138 m / s² to 1.927 m / s². With a fuzzy PID control strategy, the acceleration ranges from -1.369 m / s² to 0.925 m / s². Using the SVUFMS-PID composite control strategy of this invention, the acceleration range further converges to -0.472 m / s² to 0.358 m / s². Furthermore, compared to conventional PID control strategies, fuzzy PID control strategies, and passive suspension, the present invention achieves improvements of 77.91%, 65.52%, and 90.83% in comfort, respectively. This demonstrates that the present invention, through the synergistic effect of time delay prediction compensation and variable domain online self-tuning, can significantly reduce vehicle vibration intensity under random road disturbances and improve system robustness and ride comfort.

[0321] Furthermore, to facilitate a quantitative comparison of the comprehensive vibration reduction compensation performance of different control strategies under a single dimensional index, this embodiment selects the sprung mass longitudinal displacement, which is directly related to vehicle body posture safety, as a representative state variable, and uses its root mean square (RMS) value as a statistical evaluation index to provide a comparison of the RMS optimization rate. This is because the sprung mass displacement is the core controlled variable of the active suspension position control, and is more sensitive to the main time delay compensation effect and the online self-tuning effect of parameters. It also shows consistency with the longitudinal velocity and longitudinal acceleration in terms of their changing trends, and can concisely reflect the differences in the comprehensive performance of different control strategies under random road disturbance conditions. Table 3 below lists the RMS values ​​and optimization rates of the sprung mass displacement under different control strategies, compared to the passive suspension's 2.519 × 10⁻⁶. -2 m, conventional PID control strategy reduces the root mean square value to 7.386 × 10 -3m, with an optimization rate of 65.79%; the fuzzy PID control strategy further reduced the cost to 4.590 × 10⁻⁶ m. -3 The optimization rate was 78.74%; the SVUFMS-PID composite control strategy of this invention reduced the root mean square value to 1.498 × 10⁻⁶. -3 The optimization rate reached 93.06%, indicating that the present invention is more effective in suppressing vehicle displacement fluctuations under random road surface disturbance conditions, while taking into account both closed-loop stability and vibration reduction compensation robustness.

[0322] Table 2

[0323] In summary, through Figures 7-9 As shown in Table 2, the simulation results and data analysis comparison reveal that the SVUFMS-PID composite control strategy based on Smith prediction compensation and variable universe fuzzy self-tuning PID proposed in this invention, when applied to an electro-hydraulic servo active suspension system, can significantly compress the response amplitudes of sprung mass displacement, velocity, and acceleration under the influence of random road disturbances and parameter uncertainties. This effectively weakens the phase lag and dynamic performance degradation caused by the master time delay, and improves the timeliness of control input compensation for disturbances and the ability to maintain stability. As a result, it achieves better comprehensive control effects in key indicators such as vehicle posture stability, ride smoothness, and passenger comfort, demonstrating the comprehensive performance advantages of the control strategy of this invention compared to the comparative strategies.

[0324] The above are preferred embodiments of this application, and are not intended to limit the scope of protection of this invention. It should be noted that those skilled in the art can make several improvements without departing from the principles of this technology, and these improvements should also be considered within the scope of protection of this application.

Claims

1. A Smith-predicted compensation variable universe fuzzy PID electro-hydraulic suspension position control method, characterized in that: The method includes the following steps: Step 1: Obtain vehicle parameters and operating status data, and establish a dynamic model of a quarter-vehicle two-degree-of-freedom electro-hydraulic active suspension; Step 2: Establish a valve-controlled cylinder electro-hydraulic servo actuator model with servo valve control voltage as input and hydraulic cylinder piston displacement as output, determine the main time delay parameters of the electro-hydraulic servo control channel, and couple the actuator model with the suspension dynamics model to construct an electro-hydraulic active suspension controlled object with time delay; Step 3: Based on the controlled object with time delay, construct a Smith prediction compensation-variable universe fuzzy PID parameter self-tuning composite controller; wherein, the Smith prediction compensation structure is used to predict and compensate the main time delay, and the PID control parameters are self-tuned online through variable universe fuzzy inference to generate the servo valve control voltage to realize the electro-hydraulic active suspension closed-loop control. Step 4: Establish a random road surface excitation model and generate random road surface excitation signals. Use the random road surface excitation signals as the road surface input of the dynamic model to construct simulation excitation conditions for performance verification. Step 5: Under the random road surface excitation and external disturbance conditions, the controller parameters, fuzzy rules, and online parameter tuning rules of the universe of discourse scaling factor of the composite controller are adaptively and finely adjusted, and the simulation results are compared and analyzed to verify the control effect and robustness of the control method.

2. The method for Smith prediction compensation variable universe fuzzy PID electro-hydraulic suspension position control according to claim 1, characterized in that: To obtain an accurate dynamic description of the suspension side and provide a foundation for subsequent electro-hydraulic servo actuator modeling and control law design, the process of establishing a quarter-vehicle two-degree-of-freedom active suspension dynamic model in step one specifically includes: Acquire the state information of the vehicle's hydraulic active suspension system, the state information including state input data and state model data used to describe the system's dynamic characteristics; Based on the state model data obtained above, a two-degree-of-freedom dynamic model of the electro-hydraulic servo active suspension is established, as shown below: A dynamic model of the active suspension is established based on Newton's second law: In the above formula, F t The expression is as follows: In the suspension dynamics model, m s The sprung mass is the sum of the mass of the car body frame and the passenger; m u This indicates the unsprung mass, which is the mass of the tire assembly; B p K represents the damping coefficient of the suspension system. s Indicates the stiffness coefficient of the suspension system; x ms Indicates the vertical displacement of the sprung mass; x mu Indicates the vertical displacement of the unsprung mass; x p F represents the output displacement of the electro-hydraulic actuator system. t K1 and K2 represent the elastic force of the tire; K1 and K2 represent the elastic force coefficients under various working conditions, and neither of them is zero. This represents the random road surface excitation input. Obtain the state information of the vehicle's hydraulic active suspension system, and establish a quarter-vehicle body two-degree-of-freedom active suspension dynamic model based on the state information. The motion differential equations for the sprung mass and unsprung mass are obtained as follows. The dynamic model is then organized into a state-space form for subsequent controller design, and the state variables are defined as follows: ， ， ， The dynamic equation above can be rewritten as:

3. The Smith prediction compensation variable universe fuzzy PID electro-hydraulic suspension position control method according to claim 1, characterized in that: Based on a quarter-vehicle two-degree-of-freedom active suspension dynamics model, a coupled controlled object of the valve-controlled cylinder electro-hydraulic servo actuator and the suspension dynamics model is established. Specifically, a nonlinear dynamics model of the valve-controlled cylinder actuator is constructed, with the servo valve control voltage as input and the hydraulic cylinder piston displacement as output. The main time-delay parameters of the control channel are identified to form a controlled object containing time-delay elements. This model is used for time-delay prediction compensation by the Smith predictor and serves as the object model for online tuning of the parameters of a variable-domain fuzzy self-tuning PID controller. The valve-controlled cylinder actuator modeling process in step two includes: Based on the structural parameters and operating data of the electro-hydraulic servo system obtained in claim 1 and the two-degree-of-freedom dynamic model of the electro-hydraulic servo active suspension established in claim 2, a nonlinear dynamic model of the valve-controlled cylinder electro-hydraulic servo actuator is constructed. This yields a nonlinear state equation with piston displacement, piston velocity, and load pressure as state variables, and provides actuator force / position input for the active suspension dynamic model. The construction process includes force balance relationships, pressure dynamic equations, valve orifice flow equations, and servo valve drive relationships, which are then uniformly written into a nonlinear dynamic model, as detailed below: To model the asymmetric cylinder system, the actuator load flow equation generated by the throttling at the servo valve orifice is first constructed: Among them, C d Let ω be the flow coefficient at the throttling orifice, and ω be the area gradient of the throttling window of the slide valve. n = A1 / A2, where A1 is the piston area of ​​the rodless chamber, A2 is the piston area of ​​the rod chamber, and x... vr Where ρ is the valve core opening and P is the liquid density. s For the oil supply pressure, P L This refers to the load pressure. Construct a linearized model equation for the valve orifice load flow: Wherein, the flow gain K in the linearized model q With pressure gain K c The load flow equation is obtained by performing small-signal linearization at the operating point, which satisfies the following conditions: , When the piston rod extends, the K q With K c The specific expressions are as follows: Similarly, when the system piston rod retracts, the flow rate and pressure gain are respectively K q =K qh K c =K ch . The continuity constraints and dynamic relationships of the valve-controlled asymmetric hydraulic cylinder electro-hydraulic servo actuator under piston rod extension conditions are established. The continuity equation of the actuator load flow and the dynamic equation of the piston load system are constructed as follows: Among them, v t Let βe be the effective volume of the hydraulic system, and x be the equivalent bulk modulus of the hydraulic fluid. p Let m be the displacement of the piston extension in the system, and B be the equivalent mass. p Let F be the equivalent viscous damping coefficient, and K be the equivalent stiffness coefficient. External disturbance force F L Caused by the state variables of the suspension sprung and unsprung loads, and can be expressed as: Where A is the effective pressure-bearing area of ​​the hydraulic cylinder, in meters. u For unsprung mass, C r K is the equivalent coefficient of suspension damping. s K is the equivalent stiffness coefficient of the suspension spring. cc This represents the tire's equivalent stiffness coefficient, and x1 to x4 are suspension state variables. Establish the driving relationship between the servo valve spool displacement and the control input voltage: Assuming the servo valve spool displacement is related to the control input voltage u r The linear proportional relationship between them can be expressed as: Among them, K v This is the voltage-displacement gain coefficient of the servo valve, which is used to convert the electrical signal u r This is mapped to the adjustment amount of hydraulic flow. Based on the above valve core displacement-voltage drive relationship, the aforementioned linearized load flow model, load flow continuity equation, and piston dynamics equation, and by combining and simplifying these equations and eliminating the load flow and load pressure variables, the dynamic equation for the extension / retraction displacement of the hydraulic cylinder piston rod can be obtained. Construct the system dynamic equations under the piston rod extension condition: in, . Construct the system dynamic equations under the piston rod retraction condition: in, Construct the state differential equations of the hydraulic system: Choose piston displacement, piston velocity, and piston acceleration as state variables, and let: ， ， The state differential equations of the hydraulic system can be derived from the above system dynamic equations as follows:

4. The Smith prediction compensation variable universe fuzzy PID electro-hydraulic suspension position control method according to claim 1, characterized in that: Based on the controlled object established in claim 3, which is a valve-controlled cylinder electro-hydraulic servo actuator coupled with an active suspension dynamics model and includes master time delay parameters, in order to achieve accurate tracking of the hydraulic cylinder piston displacement and improve the robustness of the system under impact loads and parameter uncertainties, the design process of the composite controller in step three includes: Based on the structural parameters and operating data of the electro-hydraulic servo system obtained in claim 1, and the coupled controlled object model with main time delay parameters established in claims 2 and 3, a variable universe of discourse fuzzy refined adaptive PID composite control strategy with Smith prediction compensation is constructed to reduce the impact of the main time delay on closed-loop stability and tracking accuracy and to achieve online self-tuning of control parameters. Specifically, a Smith prediction structure is used to predict and compensate for the main time delay, and the PID parameters are updated online through variable universe of discourse fuzzy inference to generate the servo valve control voltage to achieve active suspension closed-loop position control. The design process of the composite control strategy includes: Construct the Smith prediction compensation structure: Based on the coupled controlled object with a main time delay as described in claim 3, a Smith prediction and compensation structure is constructed to achieve prediction and compensation for the main time delay component; wherein, the controlled object is represented as consisting of a time-delay-free part With pure delay A series-connected object model is used, and an outer loop controller is set. Simultaneously, an internal model of the Smith predictor is constructed, including a time-delay-free model. and its corresponding time delay model Based on this, the transfer function of the compensation stage is constructed: in, For the expected input, For error signals, For controller output, This is the system output. Based on the Smith prediction compensation structure, the closed-loop transfer function can be expressed as: When the internal model is dynamically consistent with the controlled object and the time delay parameters are accurately identified, that is, when the internal model is consistent with the controlled object and the time delay parameters are accurately identified, the following conditions are met. and If the actual master time delay is consistent, then the closed-loop transfer function can be equivalent to: At this point, the closed-loop characteristic equation can be equivalently expressed as: This results in the closed-loop poles being jointly determined by the time-delay-free object and the outer-loop controller, and the main time-delay term no longer entering the closed-loop characteristic polynomial; the system output only exhibits a pure delay element e. -τs Therefore, the Smith prediction compensation structure can weaken the adverse effects of the master time delay on system stability and tracking performance, and realize prediction compensation for elements containing master time delay. Constructing a universe-scaling mechanism for variable universe-of-discourse fuzzy self-tuning PID: To avoid insufficient resolution in the small error region due to a fixed initial universe of discourse, which could lead to a decrease in steady-state accuracy, a dynamic universe of discourse adjustment mechanism is introduced into the variable universe of discourse fuzzy self-tuning PID: based on the predicted output formed by the Smith prediction compensation structure, the control error is constructed. and error change rate and will , As input for fuzzy inference; let the input variable be... , The initial universes of discourse are [−E, E] and [−E], respectively. C E C The initial universe of discourse for the output variable u is [−U, U]. Without altering the shape of the membership function and the fuzzy rules, a universe of discourse scaling factor is introduced to dynamically scale the input / output universes, resulting in the scaled universes expressed as follows: in, , For the input universe of discourse scaling factor, To output the universe of discourse scaling factor; when or When the domain is large, the dynamic response capability is improved by expanding the domain of discourse. and When the size is small, the domain of discourse is contracted to improve the accuracy of steady-state control. The universe of discourse scaling factor is driven by the error magnitude and error variation trend, and the error threshold is assumed to satisfy... ,in , Assume the error rate of change threshold satisfies ,in , And set upper and lower limits for the scaling factor to meet the requirements. ,in The value ranges from 0.3 to 0.

6. The value ranges from 1.0 to 2.0; therefore, the input scaling factor can be determined according to the following piecewise linear rule: Through the aforementioned dynamic scaling rules, the input universe of discourse automatically expands when the system is in a large error phase to enhance the adjustment amplitude and response speed; as the error gradually decreases and approaches zero, the input universe of discourse automatically shrinks to improve the control resolution and steady-state accuracy in the small error region, thereby achieving a balance between dynamic and steady-state performance. Furthermore, the input linguistic variable set is preferably divided into {NB, NM, NS, ZO, PS, PM, PB}, and the membership function is preferably a triangular membership function to ensure the continuity and feasibility of fuzzy inference under dynamic scaling conditions. Constructing the rules for determining the universe of discourse scaling factor: To achieve online, refined, and adaptive regulation of fuzzy self-tuning PID controllers with variable universe of discourse, a function-based method for constructing the universe of discourse scaling factor is adopted. The selected function must satisfy the following five conditions: consistency, monotonicity, normality, duality, and zero-avoidance. The function satisfying these five conditions can be defined as follows: Where x is the input variable; λ is the function coefficient, usually λ∈(0,1). As λ increases, α(x) decreases, but the change in α(x) is more drastic, the universe of discourse compression is obvious, and the system response is fast; K is the exponential coefficient and K>0, the larger K is, the larger α(x) is. K is the coefficient of the integral term; n is the number of input variables, here taken as 2; p i It is the i-th element in the constant vector; e i β is the i-th element in the input deviation vector; β(0) is the initial value of the output variable's universe of discourse scaling factor. Using the above functional form, the universe of discourse can be expanded to enhance dynamic response when the error is large, and contracted to improve steady-state control accuracy when the error approaches zero. Based on the above principles and through multiple experiments, the input universe scaling can be set as follows: Through analysis , and After assessing the system's control effect, the universe of discourse scaling factor for the output variable was determined. The results show that... and monotony and The monotonicity is consistent. Conversely, monotony and The monotonicity is opposite. Therefore, after many experiments, the following scaling factor was chosen and can be expressed as: in: , , Position control , , The output universe of discourse scaling factor. Constructing the process of fuzzy inference, defuzzification, and control law output: Based on the fuzzification of input variables and the establishment of a fuzzy rule base, the Mamdani-type inference method is used to represent the fuzzy conditional statement "if A then B" as a fuzzy relation R from the input domain to the output domain. When the input is A... * When, output fuzzy set B * Obtained by fuzzy synthesis: in," "This is a fuzzy composition operator. Therefore, the fuzzy Ra→ε relation Mamdani is based on the definition: The membership function of the output fuzzy set is obtained by using the supremum-minimum composition rule: To obtain a definite quantity that can be used for online parameter tuning, the output fuzzy set is defuzzified; preferably, the centroid method is used to obtain a clear output u. c ,satisfy: Alternatively, a weighted average can be used in the discretization implementation: Based on the defuzzification results, the PID parameter correction ΔK is obtained respectively. p ΔK i ΔK d And it is superimposed on the PID initial parameters to achieve online updates: in, , , All are initial setpoints for PID control; based on online updates. , , Construct control laws and output control quantities It can be represented as: and the As the drive input of the valve-controlled cylinder electro-hydraulic servo actuator, the Smith prediction compensation and variable universe fuzzy self-tuning PID work together to reduce phase lag and stability degradation caused by the master time delay, achieve high-precision tracking of the desired displacement under impact load and parameter uncertainty, effectively suppress vehicle body mass vibration, and improve the system's anti-disturbance and robustness.

5. The Smith prediction compensation variable universe fuzzy PID electro-hydraulic suspension position control method according to claim 1, characterized in that: To ensure that the verification conditions of the control method closely resemble actual vehicle road conditions and to improve the credibility of the control effect demonstration, a stochastic road excitation model is constructed and a stochastic road displacement input is generated based on the quarter-vehicle electro-hydraulic active suspension dynamic model described in claim 2 and the coupled controlled object model described in claim 3. , will the As the road surface displacement input for the dynamic model, the process of establishing and generating the random road surface excitation model includes: Constructing a spatial power spectral density model for road surface roughness: The spatial power spectral density function of road surface roughness is established with spatial frequency n as the independent variable, satisfying the power function form: Where: n is the spatial frequency; n0 is the reference spatial frequency; This refers to the road surface roughness coefficient. For frequency exponents, the preferred choice is... . The definition of the spatial power spectral density of a road surface is: in: For frequency band The actual power of the included road surface power spectral density. Complete the equivalent mapping from the spatial domain to the time domain: Let the longitudinal speed of the vehicle be... And establish the correspondence between spatial frequency and temporal frequency as follows: When the vehicle is at speed During driving, the components of the road surface vertical displacement spectrum contained in the time frequency band and the spatial frequency band are the same, i.e., the frequency band... and The power of all Therefore, the time power spectral density of the road surface displacement is obtained as follows: In frequency index Under these conditions, the above time power spectral density can be further expressed as: Therefore, the time power spectral density of the vertical velocity of the road surface can be derived as follows: As can be seen from the above, the time power spectral density of the road surface vertical velocity is a constant value, possessing the same characteristics as the power spectrum of white noise. Therefore, the white noise shaping method can be used to generate random road surface displacement input. Construct a white noise shaping filter and generate random road surface displacement input: Take a Gaussian white noise signal with a unit intensity of 1 For input, road surface displacement For the output, let the frequency response function of the shaping filter be... Then the power spectral density of the road surface displacement satisfies: in: Let the power spectral density of the white noise signal satisfy the following condition: And take: . angular frequency express At that time, the power spectral density of the road surface displacement can be written as: Therefore, the frequency response function of the shaping filter is: Therefore, the transfer function of the shaping filter is obtained as follows: Therefore, the frequency domain relation satisfies: Converting the above frequency domain relationship to the time domain yields the following road surface time domain input model: Since the road surface spectrum is approximately horizontal in the low-frequency range, a lower cutoff frequency is introduced. The time power spectral density of the road surface displacement is updated as follows: Accordingly, the frequency response function of the shaping filter is updated as follows: Therefore, the time-domain model of the filtered white noise road surface is: Furthermore, to avoid introducing deviations by using time frequency for the lower cutoff frequency, the lower cutoff time frequency is... Convert to lower cutoff space frequency The relationship between the two satisfies: Based on this, the updated random time-domain input model for the road surface is obtained as follows: in: This refers to the road surface roughness coefficient. The lower cutoff spatial frequency; Reference spatial frequency; The longitudinal speed of the vehicle; This is a white noise signal with an expected value of 0. Generate random road surface conditions and couple them with the input: The random road surface displacement excitation signal As the road displacement input of the active suspension dynamics model, and coupled with the model described in claims 2 and 3 to form a random road disturbance condition, it is used to verify the closed-loop tracking performance and robustness of the control method under random road excitation conditions.

6. The electro-hydraulic servo active suspension position control method based on Smith predictor and variable universe fuzzy self-tuning PID according to claim 1, characterized in that: The method for performing online adaptive fine-tuning and comparative verification of the composite controller under random road surface excitation and external disturbance conditions in step five specifically includes: A: Under the preset random road surface level and vehicle longitudinal speed conditions, and with the superimposed external disturbance or impact load conditions, the control error and error change rate between the desired sprung mass displacement trajectory and the actual sprung mass displacement output are used as the basis for online tuning. Based on the fuzzy inference results of the variable universe of discourse, the online update rules of the PID control parameters are adaptively corrected, and the adjustment laws of the fuzzy rules and the universe of discourse scaling factor are finely adjusted so that the composite control strategy can maintain closed-loop stability under the condition of parameter uncertainty and time delay, and improve the tracking accuracy and anti-disturbance capability of the target displacement trajectory. B: Construct a comparative control strategy under the random road surface excitation and external disturbance conditions and perform simulation verification. The comparative control strategy includes at least a passive suspension no-control strategy, a PID control strategy, a fuzzy PID control strategy, and a composite control strategy of Smith predictor and variable universe fuzzy self-tuning PID. By comparing and analyzing the longitudinal displacement response, longitudinal velocity response, and longitudinal acceleration response of the sprung mass, and combining dynamic performance indicators such as overshoot, settling time, and steady-state error, the differences between the strategies are evaluated from the dimensions of driving safety, vehicle longitudinal stability, and ride comfort. This verifies the superiority of the composite control strategy in terms of vehicle vibration suppression, improved position tracking accuracy, and enhanced robustness under random road surface and disturbance conditions.

Images