Longitudinal and vertical synchronous cooperative escape control system and method for full-active suspension electric vehicle

Abstract

The application is a longitudinal and vertical synchronous cooperative escape control system and method for a full active suspension electric vehicle. The system includes a suspension actuation module receiving an escape start signal, obtaining a suspension control target, and calculating an actuation force; a wheel dynamic load observation module observing and outputting the wheel dynamic load; a wheel dynamic load dynamic stability discrimination module introducing an ideal suspension model and a timing module, comparing the wheel dynamic load output by the wheel dynamic load observation module with the ideal wheel dynamic load output by the ideal suspension model, and completing the wheel dynamic load dynamic stability discrimination; a delay compensation module replacing the wheel dynamic load state relied on in the drive torque synchronous control module through an eigenvalue function; and a drive torque synchronous control module completing the output control of the wheel torque based on a fusion controller of model predictive control and sliding mode control. The application overcomes the problem of unsynchronized drive control and actual wheel state, reduces the delay of escape control, and realizes the maximum utilization of wheel adhesion.

Description

Technical Field

[0001] The present invention belongs to the field of vehicle-by-wire intelligent chassis dynamics control, and specifically relates to a longitudinal and vertical synchronous cooperative escape control system and method for fully active suspension electric vehicles. Background Art

[0002] Vehicles are easily stuck on low-adhesion surfaces such as snow and sand. Traditional methods of getting a stuck vehicle out of a jam usually rely on manual push carts or specialized equipment, which is inefficient and cumbersome to operate.

[0003] Fully active suspension electric vehicles, with their self-adjusting suspension, offer the potential for developing autonomous vehicle escape control. To enhance these capabilities, some automakers have developed methods based on the active suspension's vibration patterns to achieve this. However, this approach often disables the vehicle's existing anti-skid function. This is because traditional anti-skid functions first monitor wheel slip using wheel speed sensors and then suppress the drive torque to prevent slip. This "slip first, then suppression" strategy can lead to a certain deviation between wheel control and actual state. This deviation can inhibit escape capability, potentially preventing the trapped vehicle from being able to escape due to a delay in increasing the drive torque when it should, or causing excessive slip when it should, ultimately resulting in low drive torque utilization.

[0004] In addition, some manufacturers combine electronic differential locks with active suspension vibration escape methods to lock the low-adhesion wheels and concentrate the driving force on the high-adhesion wheels. However, this method also has limitations. The locked wheels can no longer use the road adhesion, and the remaining drive wheels are still subject to excessive slip when trying to escape. Slipping will destroy the existing adhesion conditions and further increase the risk of getting stuck. Therefore, how to effectively improve the longitudinal and vertical coordination when getting out of trouble and maximize the use of the wheel adhesion conditions is the main problem to be solved by this application. Summary of the Invention

[0005] The object of the present invention is to provide a longitudinal and vertical synchronous cooperative escape control system and method for a fully active suspension electric vehicle.

[0006] The technical solution to achieve the purpose of the present invention is: a longitudinal and vertical synchronous cooperative escape control system for fully active suspension electric vehicles, including a suspension actuation module, a wheel dynamic load observation module, a wheel dynamic load dynamic stability judgment module, a delay compensation module and a drive torque synchronization control module;

[0007] The suspension actuation module receives the escape start signal and obtains the suspension control target based on the acquired vehicle status information. The target is the sine and cosine wheel dynamic load with fixed amplitude A and frequency F. Based on the suspension control target, it calculates the actuation force u used to track the target and complete the suspension actuation. It then outputs the actuation force u and the amplitude A and frequency F of the target tracked under the actuation force u.

[0008] Wheel dynamic load observation module: Through the vehicle status information and the actuation force u output by the suspension actuation module, the wheel dynamic load F zdelay Make observations and output;

[0009] Wheel dynamic load dynamic stability judgment module: introduces the ideal suspension model and timing module to compare the delayed wheel dynamic load output by the wheel dynamic load observation module The ideal wheel dynamic load F output by the ideal suspension model z , the dynamic stability of the wheel load is determined by calculating the variance of the eigenvalue;

[0010] Delay compensation module: receives the output of the wheel dynamic load dynamic stability judgment module when it is judged to be stable. With F z Calculate the signal and its corresponding amplitude, frequency, and deviation characteristic values With F z The delay time γ corresponding to the phase difference delay , the replacement of the wheel dynamic load state that is dependent on the drive torque synchronization control module is completed through the eigenvalue function;

[0011] Drive torque synchronization control module: A fusion controller based on model predictive control and sliding mode control completes the output control of wheel torque.

[0012] Further, the following steps are included:

[0013] Step (1): Obtain vehicle status information and determine whether to enter the longitudinal and vertical collaborative escape process;

[0014] Step (2): The suspension actuation module obtains the suspension control target, calculates and outputs the actuation force u for completing the suspension actuation and the amplitude A and frequency F of the corresponding tracking target according to the suspension control target;

[0015] Where A = α min (h u ,h d ), α is the available margin of amplitude, ranging from 0 to 1, h u , h d They are the upper and lower jump limits of the suspension respectively; k is the equivalent spring stiffness of the vehicle, and m is the mass of the vehicle;

[0016] Step (3): The wheel dynamic load observation module receives the acquired vehicle status information and the actuating force u to realize the wheel dynamic load The observation is simultaneously output to the wheel dynamic load stability judgment module and the driving torque synchronization control module;

[0017] Step (4): The ideal suspension model of the wheel dynamic load dynamic stability judgment module starts timing according to the force u in step (2) and calculates the ideal wheel dynamic load F z ; Extract ideal wheel dynamic load F z Eigenvalues ​​amplitude A, frequency F, and offset B in the time domain; extract wheel dynamic loads with delay Eigenvalue amplitude A′, frequency F′, and offset B′ in the time domain; calculate the variance S of amplitude A′, frequency F′, and offset B′ s ′, variance value S s ′ and threshold S s Compare and judge the stability; if it is stable, go to step (5) to enter the delay compensation process; if it is unstable, go to step (3);

[0018] Step (5): The delay compensation module receives the output of the wheel dynamic load dynamic stability judgment module when it is judged to be stable. With F z Calculate the signal and its corresponding amplitude, frequency, and deviation characteristic values With F z The delay time γ corresponding to the phase difference delay , for the delay time γ delay Determine the stability and output the dynamic load status

[0019] Step (6): The drive torque synchronization control module receives the output dynamic load state of step (5) And step (3) wheel dynamic load Observed value; if the result of step (4) is unstable, the output wheel dynamic load of step (3) is used The observed value is used as the state input and then goes to step (63) to calculate the driving torque T of the driving motor. d , to achieve collaborative escape control; if step (4) is determined to be stable, the output dynamic load state of step (5) is used Replace step (3) wheel dynamic load The observed value is used as the state input and then goes to step (61) to calculate the driving torque T of the driving motor. d , to achieve coordinated escape control;

[0020] Step (7): Determine whether the escape is successful. If so, end the longitudinal and vertical coordinated escape process, and the active suspension and drive motor continue to execute normal driving commands; if not, go to step (2) until the escape mode is completed.

[0021] Furthermore, step (4) specifically includes the following steps:

[0022] Step (41): The ideal suspension model of the wheel dynamic load dynamic stability judgment module starts timing according to the force u in step (2) and calculates the ideal wheel dynamic load F z ;

[0023] Step (42): According to the ideal wheel dynamic load F z , using the rolling time window to store the wheel dynamic load F within the time period T z , the offset B is obtained by analysis;

[0024] Step (43): Based on the amplitude A and frequency F output by the suspension actuation module and the offset B output by the ideal model, the ideal wheel dynamic load F after dynamic stable vibration is obtained. z :

[0025] F Z =Asin(F·t)+B;

[0026] Where t is time;

[0027] Step (44): The wheel dynamic load dynamic stability judgment module receives the wheel dynamic load Then, use the rolling time window to store the time period T Perform eigenvalue analysis to obtain amplitude A′, frequency F′, and offset B′;

[0028] Step (45): According to the amplitude A′, frequency F′ and offset B′, the wheel dynamic load with delay after dynamic stable vibration is obtained.

[0029]

[0030] Step (46): Calculate the variance value S of each characteristic value of the wheel dynamic load signal within the time period T using the variance formula s ′:

[0031]

[0032] where x i , The variable x of the same type can be A', F' or B', and its subscript i represents the discrete sampling point of the same type variable x at the time i, is the average value of A′, F′ or B′ in the time period, where n is the total number of sampling points in the time period T;

[0033] Step (47): Set the variance value S s ′ and set threshold S sIf the wheel load is smaller than the threshold, the dynamic load is considered to be stable. With F z The signal and the corresponding amplitude, frequency and deviation characteristic values ​​are input into the delay compensation module and then transferred to step (5) to enter the delay compensation process. If it is greater than the threshold, it is considered that the wheel dynamic load is dynamically unstable and then transferred to step (3), that is, the wheel dynamic load output by the wheel dynamic load observation module is used before stabilization. Perform drive control.

[0034] Furthermore, step (5) specifically includes the following steps:

[0035] Step (51): Dynamic load on the wheel with delay and the ideal wheel dynamic load F z Use the rolling time window to store and analyze the phase difference between the two in the time period T′ to obtain the delay time γ delay ;

[0036] Step (52): Calculate the delay time γ within the time period T′ window delay The variance S′ d :

[0037]

[0038] where y i , The variable y is of the same type as the delay time γ delay , where the subscript i represents the discrete sampling point of the same type variable y at the time i, is the average value of the delay time in the time period, where n is the total number of sampling points in the time period T′;

[0039] Step (53): Substitute the variance S′ of the delay time d and the set variance threshold S d Compare and judge whether the delay time output is stable. If the variance S′ d Greater than the variance threshold S d , that is, unstable, then go to step (54); if the variance S′ d Less than the variance threshold S d , that is, stable, then go to step (57);

[0040] Step (54): Delay time is not compensated, i.e. γ delay =0;

[0041] Step (55): Obtain the dynamic load state according to the amplitude A′, frequency F′ and offset B′ of step (44) for:

[0042]

[0043] Among them, t c Delay time compensation term, used to further adjust the pre-compensation time that cannot be calculated by this method;

[0044] Step (56): Delay compensation module outputs dynamic load status The delay compensation process ends and jumps to step (61);

[0045] Step (57): Compensate for the calculated delay time γ delay ;

[0046] Step (58): Use the amplitude A', frequency F' and offset B' obtained in step (44) to obtain the dynamic load state for:

[0047]

[0048] Step (59): Delay compensation module outputs dynamic load status The delay compensation process ends and jumps to step (61).

[0049] Furthermore, step (6) includes the following steps:

[0050] Step (61): The suspension vibration makes the wheel dynamic load reach dynamic stability, and the wheel dynamic load state in the control of the driving torque of the driving motor is The signal prediction uses the eigenvalue function to replace the explicit method, and the driving torque T of the driving motor is d The output is based on the fusion controller implementation:

[0051] The eigenvalue function predicts the wheel dynamic load state as follows:

[0052]

[0053] Where Δt is the sampling step size, is the predicted dynamic load value of the next sampling point at time t, and the delay time γ delay The value of is regulated by step (53);

[0054] T d for:

[0055]

[0056] in, Complete the optimization solution for the model prediction step, The solution is completed by sliding mode switching control;

[0057] Step (62): The longitudinal and vertical coordinated escape process ends and jumps to step (7);

[0058] Step (63): The suspension vibration does not make the wheel dynamic load reach dynamic stability, and the state signal of the fusion controller is used to predict the wheel dynamic load state using an explicit method. After completing the prediction step, the state is updated and the driving torque T of the driving motor in the corresponding state is obtained. d ,in Expressed as:

[0059]

[0060] Where Δt is the sampling step size, is the predicted dynamic load value for the next sampling point at time t;

[0061] T d Expressed as:

[0062]

[0063] in, Complete the optimization solution for the model prediction step, The solution is completed by sliding mode switching control;

[0064] Step (64): The longitudinal and vertical coordinated escape control process ends and jumps to step (7).

[0065] Furthermore, the driving motor torque T d The calculation is as follows:

[0066] The discrete sliding surface S in sliding mode control is calculated by model predictive control. t The prediction is solved by the optimization algorithm As part of the control output; where the sliding surface S t for:

[0067]

[0068] e t =λ t -λ d ,

[0069] where e t is the actual slip rate λ at time t t and the target slip rate λ d The error between them, η is the gain coefficient, and n is the size of the integral window after discretization. The wheel slip rate λ is discretized using the implicit Euler method, and the semi-implicit state transfer equation predicted by the model is obtained by inverse solution as follows:

[0070]

[0071] Among them, ·τ is the relevant parameter of the controlled wheel; m is the vehicle mass in kg; r is the wheel radius in m; I is the wheel moment of inertia in kg·m; ω is the wheel speed in rad / s. The wheel angular velocity at the next moment is solved using the explicit Euler method. F xij ,ij∈{fl,fr,rl,rr} are the longitudinal forces N of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively. The Dugoff tire model is used to solve F x , F x =f(F z ,C x ,μ,λ);

[0072] Solution optimization function J completed The solution:

[0073]

[0074] st

[0075] |Δu(t)|≤Δu(t) max

[0076] T min ≤u(t)≤T max

[0077] in, It is represented by the controlled motor at time t Input: P is the penalty factor of the sliding surface, Q is the penalty factor of the motor drive torque conversion degree, and E is the penalty factor of the motor drive torque conversion degree; Δu(t) max 、T min 、T max Determined by the characteristics of the drive motor;

[0078] in, The input is obtained by the following formula:

[0079]

[0080] Where ε is the switching condition boundary layer thickness, k>0 is the gain coefficient;

[0081] The final driving torque is controlled as:

[0082]

[0083] Compared with the prior art, the present invention has the following significant advantages:

[0084] 1. The longitudinal and vertical coordinated escape control method proposed in the present invention introduces an ideal suspension model in the driving torque controller as a delay comparison object of the wheel vertical load state signal; when it is observed that the wheel dynamic load changes according to a certain rule, the introduced ideal model can replace the wheel dynamic load signal in the state observer in the original control loop, and the delay between the actuator and the state observation part in the control loop can be calculated by calculating the phase difference between the ideal model and the regular dynamic load signal obtained by observation; the ideal model comparison compensation mechanism also introduces an empirically adjustable delay time item, which can realize empirical compensation for the delay introduced by the controller calculation.

[0085] 2. The compensation mechanism based on the ideal suspension dynamic load model proposed in the present invention is specifically divided into two layers. When the conditions of the first layer are met, the ideal suspension model can completely replace the original dynamic load observer. This layer introduces an adjustable item for empirical compensation of the controller calculation delay time, which can reduce the delay problem caused by the control calculation. The second layer is executed when the conditions of the first layer are met and the delay compensation time output of the calculated phase difference tends to be stable. In the second layer, the delay time caused by the sensor measurement processing and the actuator response in the control loop can be further compensated, thereby achieving high-synchronization control.

[0086] 3. This invention designs a semi-implicit state transfer equation for coordinated longitudinal and vertical drive torque control, incorporating ideal model control inputs. Specifically, the transfer equation for the controlled wheel slip is solved implicitly, while some states are implemented explicitly. The ideal model output is the delay-compensated vertical dynamic load on the wheel, which can be converted through the tire model and used to calculate the wheel slip, specifically using the following equation:

[0087]

[0088] This equation can realize the state transfer expression of the wheel slip rate. Even if the compensation mechanism mentioned above is not involved in compensation, the vertical load can still be used as the state input of the wheel slip rate state transfer equation in an explicit manner, and a certain degree of synchronous control can still be achieved.

[0089] 4. The longitudinal and vertical collaborative escape control proposed in the present invention provides a specific method that combines model predictive control with sliding mode control. This fusion method has the advantages of both the predictability of model predictive control and the high robustness of sliding mode control, and the stability of the final controller is still determined by the sliding mode controller. Compared with traditional predictive sliding mode control, the fusion method given in this patent predicts the sliding surface by means of model prediction, and completes the solution of the equivalent control law in the sliding mode control law by means of optimization solution; the switching law in the sliding mode control still plays a role in robustness and stability. BRIEF DESCRIPTION OF THE DRAWINGS

[0090] Figure 1 This is a flow chart of the escape control method for this application;

[0091] Figure 2 This is the vertical and vertical collaborative escape flow chart for this application;

[0092] Figure 3 A flowchart of delay compensation for this application;

[0093] Figure 4 This is the driving slip rate control effect diagram under the cooperative control escape control obtained in this application;

[0094] Figure 5 This is a comparison diagram of the synchronization between the wheel dynamic load and the driving torque obtained in this application. DETAILED DESCRIPTION

[0095] The technical solution of this application will be described in detail below with reference to the accompanying drawings.

[0096] The present application provides a method for coordinated escape control of longitudinal and vertical synchronization for fully active suspension electric drive vehicles. It mainly realizes more effective longitudinal and vertical coordination by "same frequency" output control of the driving torque of the driving motor after the active suspension generates regular vibrations to increase the dynamic load of the wheels, overcomes the problem of asynchrony between the driving control and the actual wheel state, reduces the driving delay in the escape control, and realizes maximum wheel adhesion utilization.

[0097] The above-mentioned longitudinal and vertical coordinated control escape method of the present application is achieved through the following technical solutions:

[0098] S1.0: If Figure 1 Obtain vehicle status information and, based on the vehicle status information, detect whether the vehicle's longitudinal acceleration and speed are continuously zero or close to zero, and whether the wheel speed is high to determine whether the vehicle is stuck; if the vehicle is stuck, a longitudinal and vertical coordinated escape start signal is generated and the longitudinal and vertical coordinated escape process is entered to the next step S2.0; otherwise, if the vehicle is not stuck, no escape start signal is generated and the longitudinal and vertical coordinated escape process is not entered, and normal control of the active suspension and drive motor continues;

[0099] The longitudinal and vertical coordinated escape process of the present invention is specifically composed of a suspension actuation module, a wheel dynamic load observation module, a wheel dynamic load dynamic stability judgment module, a delay compensation module and a drive torque synchronization control module.

[0100] Specifically, the suspension actuation module receives the escape start signal and obtains the suspension control target based on the acquired vehicle state information. The target is expressed as a sine and cosine wheel dynamic load with a fixed amplitude A and frequency F. Based on the control target in the module, the actuation force u for completing the suspension actuation and the corresponding amplitude A and frequency F for tracking the target are calculated and output.

[0101] Wheel dynamic load observation module: mainly through the obtained vehicle status information and force u, it can realize the wheel dynamic load F zdelay Observation and output.

[0102] Wheel dynamic load dynamic stability judgment module: Its main function is to analyze the F with delay characteristics output by the wheel dynamic load observation module. zdelay , and the F without delay characteristic introduced by the ideal suspension model output z , can realize the analysis of the characteristic values ​​of each periodic signal, can complete the dynamic stability judgment of the wheel dynamic load by calculating the variance of each characteristic value, and also has a timing function. Only when it is dynamically stable, will it be further With F z The corresponding amplitude, frequency and deviation characteristic values ​​enter the delay compensation module, and the delay compensation module can process the signal delay problem to achieve longitudinal and vertical coordinated escape control.

[0103] The ideal model of active suspension introduced in it can realize the dynamic load F of the wheel. z The output equation is:

[0104]

[0105] Among them, m s ,m t are the sprung mass and unsprung mass in kg; K s ,K t are the suspension spring stiffness coefficient and tire stiffness coefficient N / m respectively; C s is the suspension damping coefficient N / (m / s); F s is the force N generated by the active suspension actuator; Z s ,Z t ,Z r are the vertical displacements m of the sprung mass, unsprung mass, and ground input respectively; g is the acceleration of gravity m / s 2 In steady state, it can output the deviation of periodic eigenvalue B.

[0106] Delay compensation module: mainly receives the output from the wheel dynamic load dynamic stability judgment module when it is judged to be stable. With F z Signal and corresponding amplitude, frequency and deviation characteristic values, to calculate With F z The delay time γ corresponding to the phase difference delay , and also has the ability to judge by γ delay The stability of γ is determined by delay Output function. It can also be realized by combining amplitude, frequency and deviation characteristic value with γ delayThe eigenvalue function represented by etc. completes the replacement of the wheel dynamic load state relied on in the driving torque synchronization control module.

[0107] Drive torque synchronization control module: A fusion controller based on model predictive control and sliding mode control completes the output control of wheel torque.

[0108] S2.0: Input of vertical and longitudinal coordinated escape start signal. The specific process is as follows: Figure 2 The suspension actuation module receives the escape start signal and obtains the suspension control target based on the acquired vehicle status information. The target is expressed as a sine and cosine wheel dynamic load with a fixed amplitude A and frequency F. Based on the control target in the module, the module calculates and outputs the actuation force u for completing the suspension actuation and the amplitude A and frequency F of the corresponding tracking target. The output signal of the suspension actuation module is simultaneously input into the wheel dynamic load dynamic stability judgment module and the wheel dynamic load observation module.

[0109] In this patent, A=α·min(h u ,h d ), α is the available margin of amplitude, ranging from 0 to 1, and h u , h d are the suspension upward jump limit and downward jump limit m respectively; k is the equivalent spring stiffness of the vehicle in N / m, and m is the vehicle mass in kg;

[0110] S3.0: Further wheel dynamic load observation module receives the acquired vehicle status information and the actuating force u to realize the wheel dynamic load The observation is output to the wheel dynamic load dynamic stability judgment module, which is transferred to step S4.0 and the drive torque synchronization control module, which is output to S6.2; the main function of outputting to S6.2 is that when the wheel dynamic load is obtained When the wheel dynamic load dynamic stability judgment module fails to judge it as stable (the default judgment result is unstable), it is still necessary to use the wheel dynamic load with relatively high delay output by the wheel dynamic load observation module. Complete the control of driving torque;

[0111] S4.0: The wheel dynamic load dynamic stability judgment module receives the amplitude A, frequency F and actuation force u output by the suspension actuation module and the wheel dynamic load output by the wheel dynamic load observation module. Set the timer to t and perform the following steps:

[0112] The timing function is only activated when the actuating force u output by the suspension actuation module is received;

[0113] S4.1: Further, the received actuating force u is input into the ideal model of the active suspension to obtain the ideal wheel dynamic load F zThe rolling time window is used to store the wheel dynamic load F within a period of time T. z , the offset B is obtained by analysis. The time window size must contain at least 3 cycles of the measured signal, that is, T≥3 / F;

[0114] S4.2: The received amplitude A, frequency F, and offset B are then extracted and stored. Based on the amplitude A and frequency F output by the suspension actuation module and the ideal model output, the dynamic load after dynamic stable vibration satisfies the following relationship:

[0115] F Z =Asin(F·t)+B; (2)

[0116] S4.3: Obtain the wheel dynamic load received by the wheel dynamic stability judgment module Then, use the rolling time window to store T within a period of time Perform eigenvalue analysis to determine amplitude A′, frequency F′, and offset B′. The time window must contain at least three cycles of the measured signal, i.e., T ≥ 3 / F, which should be consistent with T in S4.1.

[0117] S4.4: Then, from the observations of the wheel dynamic loads The amplitude A', frequency F' and offset B' information of the periodic characteristic signal are extracted and stored. The dynamic load signal obtained is delayed due to a certain calculation process. The obtained data has delayed information in time. The above observer dynamic load relationship can be described as:

[0118]

[0119] S4.5: To ensure that the wheel dynamic load state output by the wheel dynamic load observation module is used to drive the torque synchronization control module, its stability must be guaranteed. The variance formula is further used to calculate the variance value S of each eigenvalue of the wheel dynamic load signal. s ′:

[0120]

[0121] where x i , The variable x of the same type can be A', F' or B', and its subscript i represents the discrete sampling point of the same type variable x at the time i, is the average value of A′, F′ or B′ in the time period, where n is the total number of sampling points in the time period T;

[0122] S4.6: Going one step further, the variance value S s ′ and set threshold S sIf the wheel load is smaller than the ideal threshold, the dynamic load is considered to be stable and With F z The signal and the corresponding amplitude, frequency and deviation characteristic values ​​are input into the delay compensation module and the process goes to step S5.0. If it is greater than the ideal threshold, it is considered that the wheel dynamic load is unstable and the process goes to step S3.0. That is, the wheel dynamic load output by the wheel dynamic load observation module is used before it is stable. Perform drive control;

[0123] S5.0: further extract the information stored in S4.2 and S4.4 and transmit it to the delay compensation module, and enter the delay compensation process of the delay compensation module, such as Figure 3 As shown, the specific process is as follows:

[0124] S5.1: What you will get With F z The signal is stored in a rolling time window, and the phase difference between the two signals during this period T′ is analyzed to obtain the corresponding delay time γ delay ; T′ = βΔt in time T′, Δt is the sampling interval, that is, β sample points are required;

[0125] S5.2: Further, calculate the variance S′ of the delay time within the time window T′ d , further ensuring the stability of the delay compensation time, and avoiding excessive or insufficient compensation that may deteriorate the vertical and longitudinal coordination:

[0126]

[0127] where y i , The variable y is of the same type as the delay time γ delay , where the subscript i represents the discrete sampling point of the same type variable y at the time i, is the average value of the delay time in the time period, where n is the total number of sampling points in the time period T′;

[0128] S5.3: Further, the variance of the delay time S′ d and set the variance threshold S d Compare and determine whether the delay time output is stable. If it is unstable, that is, the calculated variance value is greater than the set threshold, then go to step S5.4; if it is stable, that is, the calculated variance value is less than the set threshold, then go to step S5.7;

[0129] S5.4: Delay time output is unstable, delay time is not compensated, that is, γ delay =0, no compensation is made to prevent the controlled system from becoming unstable;

[0130] S5.5: Further, the wheel dynamic load is calculated using the periodic signal characteristic value obtained in step S4.4, and the dynamic load state is obtained. It can be expressed as:

[0131]

[0132] Among them, t c Delay compensation term, used to further adjust the pre-compensation time that cannot be calculated by this method.

[0133] S5.6: Delay compensation module output dynamic load status The delay compensation process ends and jumps to step S6.0;

[0134] S5.7: Delay time output is stable, compensate the calculated delay time γ delay ;

[0135] S5.8: Further, the wheel dynamic load is calculated using the periodic signal characteristic value obtained in step S4.4 to obtain the dynamic load state It can be expressed as:

[0136]

[0137] Where Δt is the sampling step, t c Delay time empirical compensation term, used to further adjust the pre-compensation time that cannot be calculated by this method.

[0138] S5.9: Delay compensation module output dynamic load status The delay compensation process ends and jumps to step S6.0;

[0139] S6.0: At this point, the suspension vibration makes the wheel dynamic load reach dynamic stability, and the wheel dynamic load state signal of the drive torque synchronization control module is replaced by the wheel dynamic load observation module input to the delay compensation module input. Specifically, the wheel dynamic load state signal in the control of the drive motor drive torque is replaced. The signal prediction can be replaced by the explicit method using the eigenvalue function, and the driving torque T of the driving motor is d The output is realized based on sliding mode predictive control. The specific steps are as follows:

[0140] The eigenvalue function predicts the wheel dynamic load state as follows:

[0141]

[0142] Where Δt is the sampling step size, is the predicted dynamic load value of the next sampling point at time t, and the delay time γ delay The value of is regulated by S5.3;

[0143] T d is represented as:

[0144]

[0145] in, Complete the optimization solution for the model prediction step, The solution is achieved by sliding mode switching control.

[0146] Drive motor torque T d The calculation is as follows:

[0147] Discrete Sliding Surface in Sliding Mode Control Using Model Predictive Control The prediction can be solved by the optimization algorithm As part of the control output, e t is the actual slip rate λ at time t t and the target slip rate λ d The error between η and n is the gain coefficient, and n is the size of the discretized integration window. The wheel slip λ is discretized using the implicit Euler method, and the semi-implicit state transfer equation predicted by the model is obtained by inverse solution as follows:

[0148]

[0149] Where ·τ represents the parameters related to the controlled wheel; m is the vehicle mass in kg; r is the wheel radius in m; I is the wheel moment of inertia in kg·m; ω is the wheel speed in rad / s. The wheel angular velocity at the next moment is solved using the explicit Euler method. F xij ,ij∈{fl,fr,rl,rr} are the longitudinal forces N of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively. The Dugoff tire model is used to solve F x , F x =f(F z ,C x ,μ,λ);

[0150] The solution to the optimization function J can be completed The solution:

[0151]

[0152] in, It is represented by the controlled motor at time t Input: P is the penalty factor of the sliding surface, Q is the penalty factor of the motor drive torque conversion degree, and E is the penalty factor of the motor drive torque conversion degree; Δu(t) max 、T min 、T max Determined by the characteristics of the drive motor.

[0153] in, The input is obtained by the following formula:

[0154]

[0155] Where ε is the switching condition boundary layer thickness, k>0 is the gain coefficient;

[0156] The final driving torque is controlled as:

[0157]

[0158] S6.1: End the longitudinal and vertical coordinated escape control sub-process and jump to step S7.0;

[0159] S6.2: Suspension vibration does not make the wheel dynamic load reach dynamic stability. The corresponding fusion controller state signal can only be used to predict the wheel dynamic load state using the explicit method. After completing the prediction step and updating the state, the driving torque T of the driving motor in the corresponding state is obtained. d ,in Expressed as:

[0160]

[0161] in, is the predicted dynamic load value for the next sampling point at time t.

[0162] T d is represented as:

[0163]

[0164] in, Complete the optimization solution for the model prediction step, The solution is achieved by sliding mode switching control.

[0165] S6.3: The longitudinal and vertical coordinated escape control process ends, and the process jumps to step S7.0;

[0166] S7.0: Determine whether the vehicle has successfully escaped based on the vehicle status. If so, end the longitudinal and vertical coordinated escape sub-process, and the active suspension and drive motor continue to execute normal driving commands; if not, go to step S2.0 until the escape is completed.

[0167] The following MATLAB / Simulink-CarSim joint simulation platform constructs the longitudinal and vertical coordinated escape process. The vehicle is stuck in place and enters the longitudinal and vertical coordinated escape mode. The vehicle is equipped with a motor-driven vehicle equipped with a fully active suspension. The drive control delay is 40ms, and t cThe experience-adjustable compensation item is set to 0, and the results only show the results related to the left front wheel of the vehicle: active suspension vibration control is intervened at 0 seconds, and drive control is intervened at 0.3 seconds (i.e., the longitudinal and vertical coordinated control is officially started). Figure 4 and Figure 5 The “with delay compensation” in the patent refers to the compensation mechanism of the high synchronization coordinated control of the longitudinal and vertical directions given in this patent; the “without delay compensation” refers to the high synchronization coordinated control of the longitudinal and vertical directions given in this patent without compensation mechanism (i.e., γ delay = 0) but still uses a semi-implicit method to predict the wheel rotation rate and the fusion controller to control the result; "ordinary drive control" refers to the result obtained by the sliding mode control commonly used in traditional drive anti-slip control;

[0168] The slip rate of the wheels during the vehicle escape process changes as follows: Figure 4 As shown, under conventional drive cooperative control, significant slip and large fluctuations occur. Using the proposed method, even without compensation, the slip rate is somewhat controlled, at least preventing excessive slip and loss of road adhesion. However, there is a tendency to suppress wheel drive, hindering escape. With compensation (i.e., the results of delayed compensation control), it can be seen that around 1.6 seconds, the state signal observed by the wheel dynamic load observer is replaced by a sinusoidal function of the characteristic value. The subsequent slip rate is further controlled and very close to the set target.

[0169] The synchronicity can be determined based on the simulation results. Figure 5 As shown, the simulation results of 1 to 3 seconds are mainly displayed, which show the vertical dynamic load changes under the actual state of the wheel and the drive output under the corresponding control. It can be found that in the initial stage, the three are almost overlapping and there is a delay, but after a period of time, the advantages of predictive control proposed in this article are reflected. Compared with ordinary drive control, the use of predictive control can improve a certain degree of synchronization. At this time, the predictive control actually adopts the key step S5.5 described in the patent application. When the key step S5.8 described in this patent application is completed, the delay compensation is obviously more synchronized with the vertical dynamic load than the one without delay compensation, and the adhesion ability is greater during the longitudinal and vertical coordinated escape control process.

[0170] The above are exemplary embodiments of the present application. The scope of protection of this application is defined by the claims and their equivalents. Those skilled in the art should understand that this application is not limited to the above examples. The above examples and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the invention. The scope of protection claimed in this invention is defined by the appended claims and their equivalents.

Claims

1. A longitudinal and vertical synchronous cooperative escape control system for fully active suspension trams, characterized by: It includes suspension actuation module, wheel dynamic load observation module, wheel dynamic load dynamic stability judgment module, delay compensation module and drive torque synchronization control module; The suspension actuation module receives the escape start signal and obtains the suspension control target based on the acquired vehicle status information. The target is the sine and cosine wheel dynamic load with fixed amplitude A and frequency F. Based on the suspension control target, it calculates the actuation force u used to track the target and complete the suspension actuation. It then outputs the actuation force u and the amplitude A and frequency F of the target tracked under the actuation force u. Wheel dynamic load observation module: Through the vehicle status information and the actuation force u output by the suspension actuation module, the wheel dynamic load is Make observations and output; Wheel dynamic load dynamic stability judgment module: introduces the ideal suspension model and timing module to compare the delayed wheel dynamic load output by the wheel dynamic load observation module The ideal wheel dynamic load F output by the ideal suspension model z , the dynamic stability of the wheel load is determined by calculating the variance of the eigenvalue; Delay compensation module: receives the output of the wheel dynamic load dynamic stability judgment module when it is judged to be stable. With F z Calculate the signal and its corresponding amplitude, frequency, and deviation characteristic values With F z The delay time γ corresponding to the phase difference delay , the replacement of the wheel dynamic load state that is dependent on the drive torque synchronization control module is completed through the eigenvalue function; Drive torque synchronization control module: A fusion controller based on model predictive control and sliding mode control completes the output control of wheel torque.

2. A control method for performing longitudinal and vertical synchronous coordinated escape based on the system of claim 1, characterized in that: The steps include: Step (1): Obtain vehicle status information and determine whether to enter the longitudinal and vertical collaborative escape process; Step (2): The suspension actuation module obtains the suspension control target, calculates and outputs the actuation force u for completing the suspension actuation and the amplitude A and frequency F of the corresponding tracking target according to the suspension control target; Where A = α min (h u ,h d ), α is the available margin of amplitude, ranging from 0 to 1, h u , h d They are the upper and lower jump limits of the suspension respectively; k is the equivalent spring stiffness of the vehicle, and m is the mass of the vehicle; Step (3): The wheel dynamic load observation module receives the acquired vehicle status information and the actuating force u to realize the wheel dynamic load The observation is simultaneously output to the wheel dynamic load stability judgment module and the driving torque synchronization control module; Step (4): The ideal suspension model of the wheel dynamic load dynamic stability judgment module starts timing according to the force u in step (2) and calculates the ideal wheel dynamic load F z ; Extract ideal wheel dynamic load F z Eigenvalues ​​amplitude A, frequency F, and offset B in the time domain; extract wheel dynamic loads with delay Eigenvalue amplitude A′, frequency F′, and offset B′ in the time domain; calculate the variance S of amplitude A′, frequency F′, and offset B′ s ′, variance value S s ′ and threshold S s Compare and judge the stability; if it is stable, go to step (5) to enter the delay compensation process; if it is unstable, go to step (3); Step (5): The delay compensation module receives the output of the wheel dynamic load dynamic stability judgment module when it is judged to be stable. With F z Calculate the signal and its corresponding amplitude, frequency, and deviation characteristic values With F z The delay time γ corresponding to the phase difference delay , for the delay time γ delay Determine the stability and output the dynamic load status Step (6): The drive torque synchronization control module receives the output dynamic load state of step (5) And step (3) wheel dynamic load Observed value; if the result of step (4) is unstable, the output wheel dynamic load of step (3) is used The observed value is used as the state input to calculate the driving torque T of the driving motor. d , to achieve collaborative escape control; if step (4) is determined to be stable, the output dynamic load state of step (5) is used Replace step (3) wheel dynamic load The observed value is used as the state input to calculate the driving torque T of the driving motor. d , to achieve coordinated escape control; Step (7): Determine whether the escape is successful. If so, end the longitudinal and vertical coordinated escape process, and the active suspension and drive motor continue to execute normal driving commands; if not, go to step (2) until the escape mode is completed.

3. The method according to claim 2, characterized in that Step (4) specifically includes the following steps: Step (41): The ideal suspension model of the wheel dynamic load dynamic stability judgment module starts timing according to the force u in step (2) and calculates the ideal wheel dynamic load F z ; Step (42): According to the ideal wheel dynamic load F z , using the rolling time window to store the wheel dynamic load F within the time period T z , the offset B is obtained by analysis; Step (43): Based on the amplitude A and frequency F output by the suspension actuation module and the offset B output by the ideal model, the ideal wheel dynamic load F after dynamic stable vibration is obtained. z : F Z =Asin(F·t)+B; Where t is time; Step (44): The wheel dynamic load dynamic stability judgment module receives the wheel dynamic load Then, use the rolling time window to store the time period T Perform eigenvalue analysis to obtain amplitude A′, frequency F′, and offset B′; Step (45): According to the amplitude A′, frequency F′ and offset B′, the wheel dynamic load with delay after dynamic stable vibration is obtained. Step (46): Calculate the variance value S of each characteristic value of the wheel dynamic load signal within the time period T using the variance formula s ′: where x i , The variable x of the same type can be A', F' or B', and its subscript i represents the discrete sampling point of the same type variable x at the time i, is the average value of A′, F′ or B′ in the time period, where n is the total number of sampling points in the time period T; Step (47): Set the variance value S s ′ and set threshold S s If the wheel load is smaller than the threshold, the dynamic load is considered to be stable. With F z The signal and the corresponding amplitude, frequency and deviation characteristic values ​​are input into the delay compensation module and then transferred to step (5) to enter the delay compensation process. If it is greater than the threshold, it is considered that the wheel dynamic load is dynamically unstable and then transferred to step (3), that is, the wheel dynamic load output by the wheel dynamic load observation module is used before stabilization. Perform drive control.

4. The method according to claim 3, characterized in that Step (5) specifically includes the following steps: Step (51): Dynamic load on the wheel with delay and the ideal wheel dynamic load F z Use the rolling time window to store and analyze the phase difference between the two in the time period T′ to obtain the delay time γ delay ; Step (52): Calculate the delay time γ within the time period T′ window delay The variance S′ d : where y i , The variable y is of the same type as the delay time γ delay , where the subscript i represents the discrete sampling point of the same type variable y at the time i, is the average value of the delay time in the time period, where n is the total number of sampling points in the time period T′; Step (53): Substitute the variance S′ of the delay time d and the set variance threshold S d Compare and judge whether the delay time output is stable. If the variance S′ d Greater than the variance threshold S d , that is, unstable, then go to step (54); if the variance S′ d Less than the variance threshold S d , that is, stable, then go to step (57); Step (54): Delay time is not compensated, i.e. γ delay =0; Step (55): Obtain the dynamic load state according to the amplitude A′, frequency F′ and offset B′ of step (44) for: Among them, t c Delay time compensation term, used to further adjust the pre-compensation time that cannot be calculated by this method; Step (56): Delay compensation module outputs dynamic load status The delay compensation process ends and jumps to step (61); Step (57): Compensate for the calculated delay time γ delay ; Step (58): Use the amplitude A', frequency F' and offset B' obtained in step (44) to obtain the dynamic load state for: Step (59): Delay compensation module outputs dynamic load status The delay compensation process ends and jumps to step (61).

5. The method according to claim 4, characterized in that Step (6) includes the following steps: Step (61): The suspension vibration makes the wheel dynamic load reach dynamic stability, and the wheel dynamic load state in the control of the driving torque of the driving motor is The signal prediction uses the eigenvalue function to replace the explicit method, and the driving torque T of the driving motor is d The output is based on the fusion controller implementation: The eigenvalue function predicts the wheel dynamic load state as follows: Where Δt is the sampling step size, is the predicted dynamic load value of the next sampling point at time t, and the delay time γ delay The value of is regulated by step (53); T d for: in, Complete the optimization solution for the model prediction step, The solution is completed by sliding mode switching control; Step (62): The longitudinal and vertical coordinated escape process ends and jumps to step (7); Step (63): The suspension vibration does not make the wheel dynamic load reach dynamic stability, and the state signal of the fusion controller is used to predict the wheel dynamic load state using an explicit method. After completing the prediction step, the state is updated and the driving torque T of the driving motor in the corresponding state is obtained. d ,in Expressed as: Where Δt is the sampling step size, is the predicted dynamic load value for the next sampling point at time t; T d Expressed as: in, Complete the optimization solution for the model prediction step, The solution is completed by sliding mode switching control; Step (64): The longitudinal and vertical coordinated escape control process ends and jumps to step (7).

6. The method according to claim 5, characterized in that Drive motor torque T d The calculation is as follows: The discrete sliding surface S in sliding mode control is calculated by model predictive control. t The prediction is solved by the optimization algorithm As part of the control output; where the sliding surface S t for: e t =λ t -l d , where e t is the actual slip rate λ at time t t and the target slip rate λ d The error between them, η is the gain coefficient, and n is the size of the integral window after discretization. The wheel slip rate λ is discretized using the implicit Euler method, and the semi-implicit state transfer equation predicted by the model is obtained by inverse solution as follows: in,· τ are the parameters related to the controlled wheel; m is the vehicle mass in kg; r is the wheel radius in m; I is the wheel moment of inertia in kg·m; ω is the wheel speed in rad / s. The wheel angular velocity at the next moment is solved using the explicit Euler method. F xij ,ij∈{fl,fr,rl,rr} are the longitudinal forces N of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively. The Dugoff tire model is used to solve F x , F x =f(F z ,C x ,μ,λ); Optimization function J completed The solution: st |Δu(t)|≤Δu(t) max T min ≤u(t)≤T max in, It is represented by the controlled motor at time t Input: P is the penalty factor of the sliding surface, Q is the penalty factor of the motor drive torque conversion degree, and E is the penalty factor of the motor drive torque conversion degree; Δu(t) max 、T min 、T max Determined by the characteristics of the drive motor; in, The input is obtained by the following formula: Where ε is the switching condition boundary layer thickness, k>0 is the gain coefficient; The final driving torque control is:

Images