Neural network FOPID vehicle speed tracking control method based on dynamic parameter adjustment

Abstract

The invention relates to a neural network FOPID vehicle speed tracking control method based on dynamic parameter adjustment. The method comprises the steps of designing a vehicle model; wherein the vehicle model comprises a vehicle dynamics model and a motor dynamics model; an FOPID controller is designed; improving a sparrow search algorithm to obtain an improved sparrow search algorithm; optimizing the control parameters of the FOPID controller by using the improved sparrow search algorithm to obtain the optimized control parameters of the FOPID controller; and performing control parameter updating on the optimized control parameters of the FOPID controller by using the neural network model. The control parameters of the FOPID controller are optimized by using the improved sparrow search algorithm, so that the FOPID controller is more excellent in performance in an actual system, then the optimized control parameters are updated by using the neural network model, and the vehicle model is controlled by using the updated control parameters, so that tracking control of the vehicle speed is realized.

Description

Technical Field

[0001] The present application relates to the technical field of electric vehicle speed control, and in particular to a neural network FOPID vehicle speed tracking control method based on dynamic parameter adjustment. Background Art

[0002] Research and development trends in the field of electric vehicles (EVs) continue to gain traction in the automotive industry, primarily due to the increasing demand for technological development, sustainable energy utilization, and stringent regulations for environmental safety. Effective motors and sophisticated control techniques are considered essential for the rapid development of EV technology. The reliability and durability of EV systems play an important role in determining the overall effectiveness of the entire EV system. In EVs, DC motors are one of the most commonly used actuators, with key advantages such as high reliability, low cost, and easy maintenance, making them widely used in various applications.

[0003] In the speed control of electric vehicles, many controllers are used, such as PID, fuzzy logic controllers, and neural networks. Although the traditional PID controller is widely used due to its simplicity, it performs poorly in high-order, nonlinear, and complex systems. The fractional-order proportional-integral-derivative (FOPID) controller has become the industry standard controller, replacing the traditional PID controller with better performance indicators and helps to better resist disturbance control. The FOPID controller can be adjusted using rule-based, analytical, and numerical methods. Common numerical methods include particle swarm optimization (PSO), ant colony algorithm (ACO), and firefly algorithm, etc., which are used to adjust the parameters of the FOPID controller.

[0004] Since EV systems are highly time-varying, nonlinear, and uncertain, affected by road conditions, motor parameters, and external disturbances, controllers are needed to improve performance or achieve desired performance under external disturbances and handle uncertainties with minimal controller workload. Previous studies on EV speed control concluded that PID / FOPID controllers alone do not perform effectively and ensure the required dynamic performance under different EV conditions.

[0005] Therefore, it is necessary to provide a new technical solution to improve one or more problems existing in the above solutions.

[0006] It should be noted that the information disclosed in the above background technology section is only used to enhance the understanding of the background of the present application, and therefore may include information that does not constitute the prior art known to ordinary technicians in the field. Summary of the invention

[0007] The purpose of this application is to provide a neural network FOPID vehicle speed tracking control method based on dynamic parameter adjustment, thereby overcoming one or more problems caused by the limitations and defects of related technologies at least to a certain extent.

[0008] According to an embodiment of the present application, a neural network FOPID vehicle speed tracking control method based on dynamic parameter adjustment is provided, the method comprising:

[0009] Designing a vehicle model; wherein the vehicle model includes a vehicle dynamics model and a motor dynamics model;

[0010] Design FOPID controller;

[0011] Improving the sparrow search algorithm to obtain the improved sparrow search algorithm;

[0012] The control parameters of the FOPID controller are optimized by using the improved sparrow search algorithm to obtain the optimized control parameters of the FOPID controller;

[0013] Using a neural network model to update the control parameters of the optimized FOPID controller to obtain the updated control parameters;

[0014] The vehicle model is controlled by the updated control parameters to achieve tracking control of the vehicle speed.

[0015] In the embodiment of the present application, the vehicle dynamics model is:

[0016] F=0.5ρAC d (V wheel +V wind ) 2 +mgsin(β)+mgR r cos(β)+ma(1);

[0017] In the formula, ρ represents air density, A represents vehicle area, C d is the aerodynamic drag coefficient, V wheel represents the longitudinal velocity of the vehicle, V wind represents wind speed, g represents gravitational acceleration, β represents road inclination, m represents vehicle weight, and a represents vehicle acceleration;

[0018] The motor dynamics model is:

[0019]

[0020]

[0021] In the formula, R r is the rolling resistance coefficient, i represents the armature current, L a Represents the armature inductance, L f represents the winding inductance, u(t) represents the control output, R a Represents the armature resistance, R f Indicates winding resistance, L af represents mutual inductance, ω represents motor angular velocity, J represents motor moment of inertia, B represents viscosity coefficient, r represents tire radius, v represents vehicle speed, G represents transmission ratio, represents the derivative of the armature current, It represents the derivation of the motor angular velocity ω.

[0022] In the embodiment of the present application, the differential equation expression of the FOPID controller is as follows:

[0023] u(t)=K p e(t)+K i D -λ e(t)+K d D μ e(t)(4);

[0024] In the formula, u(t) represents the control output, e(t) represents the system error, λ, μ, K p , K i , K d They represent the control parameters of the controller, D represents the fractional-order differential operator;

[0025] After Laplace transform, the transfer function expression of FOPID controller is as follows:

[0026]

[0027] Where s represents the complex frequency variable in the Laplace transform, which is used to convert the differential equation in the time domain into an algebraic expression in the complex frequency domain.

[0028] In an embodiment of the present application, the step of improving the sparrow search algorithm to obtain the improved sparrow search algorithm includes:

[0029] In the process of foraging, the sparrow population is divided into three roles: explorers, followers and early warning;

[0030] Assuming there are N sparrows in the group, the sparrow population is expressed as:

[0031]

[0032] In the formula, d represents the dimension of the sparrow population, n represents the number of sparrow populations, and x n,d represents the sparrow of the nth row and dth dimension;

[0033] The fitness function of the sparrow population is calculated according to the sparrow population, and the fitness function expression is as follows:

[0034]

[0035] In the formula, f(x n,1 ,x n,2 …x n,d ) represents the fitness value of each sparrow in different dimensions;

[0036] The explorer's position update formula is as follows:

[0037]

[0038] In the formula, t represents the current iteration number, T max Indicates setting the maximum number of iterations. represents the fitness value of the qth sparrow in the d-dimensional environment when the explorer has t iterations, α∈(0,1], R2(0,1], Q represents a random number that obeys a normal distribution, L is a matrix whose elements are all 1, ST is the warning threshold, R2<ST, indicating that no predator is found and foraging activities can be performed; R2≥ST, indicating that an invader appears and the sparrow population flies to other foraging areas;

[0039] The follower's position update formula is as follows:

[0040]

[0041] In the formula, represents the best position record in the sparrow population, X worst Indicates the worst position record, A + is a matrix whose elements are all randomly assigned values ​​of 1 or -1. It represents the fitness value of the qth sparrow in the d-dimensional environment when the number of iterations of the follower is t;

[0042] The position update formula of the alerter is as follows:

[0043]

[0044] In the formula, is the global optimal position record of the sparrow population, represents the global worst position record of the sparrow population, β is a random number that follows a normal distribution, K∈[-1,1], f q is the current individual fitness value, f g Indicates the fitness value of the best position in the current area, f w Represents the fitness value of the worst position in the current area, ε is a natural constant, It represents the fitness value of the qth sparrow in the d-dimensional environment when the number of iterations of the early warning is t.

[0045] In an embodiment of the present application, the step of improving the sparrow search algorithm to obtain the improved sparrow search algorithm includes:

[0046] The improved explorer position update formula is as follows:

[0047]

[0048] In the formula, a1, a2 represent the individual learning parameters of the explorer, and C represents the dynamic parameter.

[0049] In an embodiment of the present application, the step of optimizing the control parameters of the FOPID controller by using the improved sparrow search algorithm to obtain the optimized control parameters of the FOPID controller includes:

[0050] The improved fitness function of the sparrow population algorithm is expressed as follows:

[0051]

[0052] In the formula, b1, b2, b3, and b4 represent the weights of each part of the evaluation, e(t) is the system error, and t s Adjust the time for the system.

[0053] In an embodiment of the present application, the step of updating the control parameters of the optimized FOPID controller using a neural network model to obtain the updated control parameters includes:

[0054] The neural network model is associated with the FOPID controller to obtain a neural network-based FOPID controller model, so as to update the control parameters of the optimized FOPID controller; wherein the expression of the neural network-based FOPID controller model is as follows:

[0055]

[0056] In the formula, referring to Formula 4, α represents the fractional order, corresponding to λ in the FOPID controller, and the gradient in the neural network Corresponding to the system error e(t) in the FOPID controller, η represents the learning rate, corresponding to K in the FOPID controller p , δ corresponds to K in the FOPID controller i , x t Corresponding to u(t) in FOPID controller, K dRepresents the derivative gain of the control parameter.

[0057] In an embodiment of the present application, the step of updating the control parameters of the optimized FOPID controller using a neural network model to obtain the updated control parameters includes:

[0058] set up

[0059] The neural network weight and bias update formulas for the neural network-based FOPID controller model are as follows:

[0060]

[0061] In the formula, t is the current iteration number, t+1 represents the next iteration, represents the neural network weight at iteration t, Represents the bias of the neural network at iteration t, l+1 represents the l+1th layer in the neural network, η represents the learning rate, and α is the fractional order.

[0062] In an embodiment of the present application, the step of updating the control parameters of the optimized FOPID controller using a neural network model to obtain the updated control parameters includes:

[0063] According to formula 13, the neural network weight and bias update formula of the neural network-based FOPID controller model is as follows:

[0064]

[0065] In an embodiment of the present application, the step of controlling the vehicle model by the updated control parameters to achieve tracking control of the vehicle speed includes:

[0066] The control output is adjusted according to the updated control parameter, and the control output is input into the vehicle model to control the vehicle speed.

[0067] The technical solution provided by the embodiments of the present application may have the following beneficial effects:

[0068] In one embodiment of the present application, through the above method, a FOPID controller is designed, and the control parameters of the FOPID controller are optimized and designed using an improved sparrow search algorithm to obtain the optimized control parameters of the FOPID controller, so that the FOPID controller performs better in an actual system and improves the stability and robustness of the system. Then, the optimized control parameters are updated using a neural network model to update the optimized control parameters, thereby controlling the vehicle model using the updated control parameters to achieve tracking control of the vehicle speed.

[0069] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0070] The drawings herein are incorporated into the specification and constitute a part of the specification, illustrate embodiments consistent with the present application, and together with the specification are used to explain the principles of the present application. Obviously, the drawings described below are only some embodiments of the present application, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0071] Figure 1 A flowchart schematically shows a step flow of a neural network FOPID vehicle speed tracking control method based on dynamic parameter adjustment in an exemplary embodiment of the present application;

[0072] Figure 2 The block diagram of the neural network FOPID speed tracking control system based on dynamic parameter adjustment in an exemplary embodiment of the present application is schematically shown. DETAILED DESCRIPTION

[0073] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these embodiments are provided so that this application will be more comprehensive and complete and fully convey the concepts of the example embodiments to those skilled in the art. The described features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.

[0074] In addition, the accompanying drawings are only schematic illustrations of the present application and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and thus their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. These functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor devices and / or microcontroller devices.

[0075] This example embodiment provides a neural network FOPID vehicle speed tracking control method based on dynamic parameter adjustment. Figure 1 As shown in , the method includes: steps S101 to S106.

[0076] Wherein, step S101: designing a vehicle model; wherein the vehicle model includes a vehicle dynamics model and a motor dynamics model.

[0077] Step S102: Design a FOPID controller.

[0078] Step S103: improving the sparrow search algorithm to obtain an improved sparrow search algorithm.

[0079] Step S104: Optimize the control parameters of the FOPID controller using the improved sparrow search algorithm to obtain the optimized control parameters of the FOPID controller.

[0080] Step S105: using the neural network model to update the control parameters of the optimized FOPID controller to obtain updated control parameters.

[0081] Step S106: Control the vehicle model through the updated control parameters to achieve tracking control of the vehicle speed.

[0082] In one embodiment of the present application, through the above method, a FOPID controller is designed, and the control parameters of the FOPID controller are optimized and designed using an improved sparrow search algorithm to obtain the optimized control parameters of the FOPID controller, so that the FOPID controller performs better in an actual system and improves the stability and robustness of the system. Then, the optimized control parameters are updated using a neural network model to update the optimized control parameters, thereby controlling the vehicle model using the updated control parameters to achieve tracking control of the vehicle speed.

[0083] Next, we will refer to Figure 1 to Figure 2 Each step of the above method in this example implementation is described in more detail.

[0084] In step S101, the vehicle model (ie, the electric vehicle dynamics model) is designed mainly from two aspects: the vehicle dynamics model and the motor dynamics model.

[0085] Vehicle dynamics model design: During vehicle driving, under different road conditions, the overall movement of the vehicle will be affected by some external forces, which can directly or indirectly affect the driving state of the vehicle. During vehicle driving, the total forces acting on the vehicle mainly include air resistance, slope force, rolling resistance and vehicle traction. Therefore, the vehicle dynamics model (i.e. the force conditions of the vehicle) during driving is as follows:

[0086] F=0.5ρAC d (V wheel +V wind ) 2 +mgsin(β)+mgR r cos(β)+ma(1);

[0087] In the formula, ρ represents air density, A represents vehicle area, C d is the aerodynamic drag coefficient, V wheel represents the longitudinal velocity of the vehicle, V wind represents wind speed, g represents gravitational acceleration, β represents road inclination, m represents vehicle weight, and a represents vehicle acceleration.

[0088] Vehicle motor dynamics design: The torque T of the drive motor is related to the total force F. Among them, G represents the transmission ratio and r represents the tire radius.

[0089] This application uses a DC motor to drive an electric vehicle, so the motor dynamics model is designed as follows:

[0090]

[0091] Furthermore, according to Formula 1, Formula 2, Formula 18 and Formula 19, the motor dynamics model is as follows:

[0092]

[0093] In the formula, R r is the rolling resistance coefficient, i represents the armature current, L a Represents the armature inductance, L f represents the winding inductance, u(t) represents the control output, R a Represents the armature resistance, R f Indicates winding resistance, L af represents mutual inductance, ω represents motor angular velocity, J represents motor moment of inertia, B represents viscosity coefficient, r represents tire radius, v represents vehicle speed, G represents transmission ratio, represents the derivative of the armature current, It represents the derivation of the motor angular velocity ω.

[0094] The vehicle model defines the dynamic behavior of an electric vehicle (i.e., vehicle) under different driving conditions, including vehicle dynamics models and motor dynamics models. These models describe the response of the vehicle under acceleration, deceleration, and external disturbances, and explain the relationship between its speed and control input.

[0095] Based on the vehicle model, FOPID is designed to accurately control the vehicle speed. Compared with traditional PID, FOPID controller has stronger dynamic adaptability and anti-interference ability, and can better match the complex dynamic characteristics of electric vehicles. The dynamic characteristics of vehicles are often nonlinear, time-varying and uncertain, which puts higher requirements on the controller, and the fractional integral and differential characteristics of FOPID are suitable for handling such complexity.

[0096] In step S102, the vehicle model is designed based on the actual physical characteristics of the electric vehicle, including the vehicle body, drive motor, transmission system, and interaction with the outside world (such as resistance, slope force). In this model, the input is the driving force or motor torque provided by the controller, and the output is the speed or acceleration of the vehicle. Therefore, the input-output relationship of the vehicle model directly corresponds to the output-input of the FOPID controller.

[0097] Furthermore, the goal of the FOPID controller design is to track the target speed or adjust the acceleration so that the actual speed is close to the set value. The controller will continuously generate a control signal based on the speed error and use it as the input of the vehicle model. The vehicle model generates a new speed value based on this input, and this process forms a closed-loop feedback control.

[0098] It can be understood that the FOPID controller adds two adjustable quantity modules, the integral order and the differential order, to the traditional PID controller, and improves the controller performance through these two adjustable parameter modules.

[0099] The differential equation expression based on FOPID controller is as follows:

[0100] u(t)=K p e(t)+K i D -λ e(t)+K d D μ e(t)(4);

[0101] In the formula, u(t) represents the control output, e(t) represents the system error, λ, μ, K p , K i , K d They represent the control parameters of the controller, D represents the fractional-order differential operator;

[0102] After Laplace transform, the transfer function expression of FOPID controller is as follows:

[0103]

[0104] Where s represents the complex frequency variable in the Laplace transform, which is used to convert the differential equation in the time domain into an algebraic expression in the complex frequency domain.

[0105] It should be noted that λ, μ, K p , K i , K d are the five control parameters of the controller. Among these five control parameters, λ is the integral exponential control parameter, μ is the differential exponential control parameter, and K p is the proportional control parameter of the controller, K i is the integral control parameter of the controller, K d is the differential control parameter of the controller.

[0106] In step S103, the sparrow search algorithm is improved to obtain the improved sparrow search algorithm, which mainly includes step S1031 and step S1032.

[0107] Step S1031: Design of sparrow search algorithm.

[0108] Sparrow Search Algorithm (SSA) is a swarm intelligence optimization algorithm and a method for calculating the approximate optimal solution. Inspired by the foraging behavior of sparrows, it divides the sparrow group into three roles during the foraging process: explorers, followers, and early warning.

[0109] Assuming there are N sparrows in the group, the sparrow population is expressed as:

[0110]

[0111] In the formula, d represents the dimension of the sparrow population, n represents the number of sparrow populations, and x n,d represents the sparrow of the nth row and dth dimension;

[0112] The fitness function of the sparrow population is calculated based on the sparrow population. The fitness function expression is as follows:

[0113]

[0114] In the formula, f(x n,1 ,x n,2 …x n,d ) represents the fitness value of each sparrow in different dimensions;

[0115] The explorer's position update formula is as follows:

[0116]

[0117] Where t represents the current iteration number. max Indicates setting the maximum number of iterations. It represents the fitness value of the qth sparrow in the d-dimensional environment when the explorer has t iterations. max represents the maximum number of iterations, α∈(0,1], R2(0,1], Q represents a random number that follows a normal distribution, L is a matrix whose elements are all 1, ST is the warning threshold, R2<ST, indicating that no predator is found and foraging activities can be performed; R2≥ST, indicating that an invader appears and the sparrow population flies to other foraging areas;

[0118] The follower searches for food with the explorer and competes with the explorer for food to increase the predation rate. The follower's position update formula is as follows:

[0119]

[0120] In the formula, represents the best position record in the sparrow population, X worst Indicates the worst position record, A + is a matrix whose elements are all randomly assigned values ​​of 1 or -1. It represents the fitness value of the qth sparrow in the d-dimensional environment when the number of iterations of the follower is t;

[0121] The position update formula of the alerter is as follows:

[0122]

[0123] In the formula, is the global optimal position record of the sparrow population, represents the global worst position record of the sparrow population, β is a random number that follows a normal distribution, K∈[-1,1], f q is the current individual fitness value, f g Indicates the fitness value of the best position in the current area, f w Represents the fitness value of the worst position in the current area, ε is a natural constant, It represents the fitness value of the qth sparrow in the d-dimensional environment when the number of iterations of the early warning is t.

[0124] Step S1031: Optimization of the sparrow search algorithm.

[0125] The explorer position update strategy in the sparrow search algorithm is difficult to effectively jump out of the local optimum in some cases, especially in complex or multi-peak problems. The explorer often stagnates due to the attraction of the local optimal point, making it difficult for the sparrow search algorithm to continue to explore the global optimal solution.

[0126] In order to improve the explorer's position update, an individual feedback mechanism is introduced, which combines the global optimal value and the individual historical optimal value. The explorer is allowed to balance the global optimal individual position and its own historical individual optimal value, thereby enhancing information sharing within the sparrow population and improving the algorithm's search speed and optimization accuracy. Combined with formula 8, the improved explorer position update formula is as follows:

[0127]

[0128]

[0129] In the formula, a1, a2 represent the individual learning parameters of the explorer, and C represents the dynamic parameter.

[0130] Understandably, T max Indicates the maximum number of iterations. Therefore, in the early stage of iteration, the dynamic parameter C takes a larger value, which is convenient for the sparrow search algorithm to perform global search. In the later stage of iteration, the dynamic parameter C is smaller, which is conducive to the local search ability of the sparrow search algorithm and accelerates the convergence of the algorithm.

[0131] The five key control parameters λ, μ, K of the FOPID controller are calculated offline using the sparrow search algorithm. p , K i , K d To meet the basic speed control requirements of the system, the sparrow search algorithm can perform a global search in the parameter space to find the initial optimal parameters that meet the vehicle model. These parameters serve as the basic configuration of the controller to ensure that the FOPID controller has basic control capabilities.

[0132] In step S104, the control parameters of the FOPID controller are optimized by using the improved sparrow search algorithm, and the step of obtaining the optimized control parameters of the FOPID controller includes step S1041.

[0133] In step S1041, the fitness function of the improved sparrow population algorithm is expressed as follows:

[0134]

[0135] In the formula, b1, b2, b3, and b4 represent the weights of each part of the evaluation, e(t) is the system error, and t s Adjust the time for the system.

[0136] It can be understood that the specific implementation steps of optimizing the control parameters of the FOPID controller using the improved sparrow search algorithm include the following:

[0137] 1) Initialize the parameters of the improved sparrow search algorithm including: population size, maximum number of iterations, search space range and population location;

[0138] 2) Calculate the fitness value of the individual sparrow according to formula 12, that is, the absolute value error and time integral index value of the system corresponding to different parameter values;

[0139] 3) Compare the current individual sparrow’s fitness value with the individual sparrow’s optimal fitness value, select the optimal solution as the individual’s optimal fitness value, and record the corresponding position;

[0140] 4) Compare the current individual fitness value of the sparrow with the global optimal fitness value and record the optimal solution;

[0141] 5) Update the sparrow position according to formula 9 to formula 12;

[0142] 6) Determine whether the iteration end constraint condition is met. If so, exit; otherwise, return to step 2 to continue iterative execution.

[0143] In step S105, the improved sparrow search algorithm is used to tune the five key control parameters λ, μ, K of the FOPID controller. p , K i , K d ,The improved sparrow search algorithm is optimized through iterative optimization, and the control parameters of the FOPID controller are optimized (i.e., the FOPID controller has the best combination of control parameters), which improves its control effect and anti-interference ability to cope with the vehicle model.

[0144] Furthermore, after determining the control parameters of the optimized FOPID controller, the adaptive ability of the neural network model is combined to realize dynamic parameter adjustment. The neural network model uses the stochastic gradient descent and momentum gradient descent mechanisms to update the control parameters of the FOPID controller, so that it can respond to external disturbances in real time and improve the adaptability and robustness of speed control.

[0145] In the neural network model, the target expected value calculated by the forward propagation in the network and the error value of the actual output axis are used as the loss function. The error gradient of the weight is obtained by back propagation, and the control parameters are updated.

[0146] The control parameters of the optimized FOPID controller are passed to the neural network model for the initialization phase. The neural network model uses these parameters as initial weights and biases, so that it starts working from the optimized configuration, improving the starting efficiency of the neural network model.

[0147] Furthermore, both the neural network model and the FOPID controller adjust the system and network according to the error between the expected and actual outputs, where the feedback process in the FOPID controller corresponds to the back propagation in the neural network model. Therefore, the system error e(t) in the FOPID controller can be compared with the gradient of the neural network. Corresponding associations.

[0148] The step of optimizing the control parameters of the FOPID controller by using the improved sparrow search algorithm to obtain the optimized control parameters of the FOPID controller includes step S1051, step S1052 and step S1053.

[0149] In step S1051, the neural network model is associated with the FOPID controller to obtain a neural network-based FOPID controller model to update the control parameters of the optimized FOPID controller; wherein the expression of the neural network-based FOPID controller model is as follows:

[0150]

[0151] In the formula, referring to Formula 4, α represents the fractional order, corresponding to λ in the FOPID controller, and the gradient in the neural network Corresponding to the system error e(t) in the FOPID controller, η represents the learning rate, corresponding to K in the FOPID controller p , δ corresponds to K in the FOPID controller i , x t Corresponding to u(t) in the FOPID controller.

[0152] In step S1052, set

[0153] The neural network weight and bias update formulas for the neural network-based FOPID controller model are as follows:

[0154]

[0155] In the formula, t is the current iteration number, t+1 represents the next iteration, represents the neural network weight at iteration t, Represents the bias of the neural network at iteration t, l+1 represents the l+1th layer in the neural network, η represents the learning rate, and α is the fractional order.

[0156] In step S1053, the neural network weight and bias update formula of the neural network-based FOPID controller model is obtained according to formula 13 as follows:

[0157]

[0158] In step S106, the step of controlling the vehicle model by using the updated control parameters to achieve tracking control of the vehicle speed includes step S1061.

[0159] In step S1061, the control output is adjusted according to the updated control parameters, and the control output is input to the vehicle model to control the vehicle speed.

[0160] It is understandable that in the actual operation of the system, the vehicle speed and external environment will continue to change. At this time, the neural network evaluates the input error and learns the deviation between the current vehicle speed and the target speed in real time to dynamically adjust the control parameters of the FOPID controller. The neural network uses the control parameters optimized by the improved sparrow search algorithm as a benchmark, and fine-tunes the control parameters of the FOIPD controller through the mechanism of stochastic gradient descent and momentum gradient descent to better adapt it to the current driving state.

[0161] In the feedback loop of the control system, the FOPID controller outputs a control signal (i.e., control output) based on the real-time updated control parameters to adjust the vehicle's drive motor. This control signal affects the vehicle speed through feedback control, thereby achieving precise speed control. As the neural network continues to adjust online, the FOPID controller is gradually optimized to maintain stable speed control performance under dynamic conditions.

[0162] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary technical means in the art that are not disclosed in the present application.

Claims

1. A neural network FOPID vehicle speed tracking control method based on dynamic parameter adjustment, characterized in that: The method includes: Designing a vehicle model; wherein the vehicle model includes a vehicle dynamics model and a motor dynamics model; Design FOPID controller; Improving the sparrow search algorithm to obtain the improved sparrow search algorithm; The control parameters of the FOPID controller are optimized by using the improved sparrow search algorithm to obtain the optimized control parameters of the FOPID controller; Using a neural network model to update the control parameters of the optimized FOPID controller to obtain the updated control parameters; The vehicle model is controlled by the updated control parameters to achieve tracking control of the vehicle speed.

2. The vehicle speed tracking control method based on the dynamic parameter adjustment neural network FOPID according to claim 1 is characterized in that: The vehicle dynamics model is: F=0.5ρAC d (V wheel +V wind ) 2 +mgsin(β)+mgR r cos(β)+ma(1); In the formula, ρ represents air density, A represents vehicle area, C d is the aerodynamic drag coefficient, V wheel represents the longitudinal velocity of the vehicle, V wind represents wind speed, g represents gravitational acceleration, β represents road inclination, m represents vehicle weight, and a represents vehicle acceleration; The motor dynamics model is: In the formula, R r is the rolling resistance coefficient, i represents the armature current, L a Represents the armature inductance, L f represents the winding inductance, u(t) represents the control output, R a Represents the armature resistance, R f Indicates winding resistance, L af represents mutual inductance, ω represents motor angular velocity, J represents motor moment of inertia, B represents viscosity coefficient, r represents tire radius, v represents vehicle speed, G represents transmission ratio, represents the derivative of the armature current, It represents the derivation of the motor angular velocity ω.

3. The vehicle speed tracking control method based on the dynamic parameter adjustment neural network FOPID according to claim 2 is characterized in that: The differential equation expression of the FOPID controller is as follows: u(t)=K p e(t)+K i D -λ e(t)+K d D μ e(t)(4); In the formula, u(t) represents the control output, e(t) represents the system error, λ, μ, K p , K i , K d They represent the control parameters of the controller, D represents the fractional-order differential operator; After Laplace transform, the transfer function expression of FOPID controller is as follows: Where s represents the complex frequency variable in the Laplace transform, which is used to convert the differential equation in the time domain into an algebraic expression in the complex frequency domain.

4. The vehicle speed tracking control method based on the dynamic parameter adjustment neural network FOPID according to claim 3 is characterized in that: The step of improving the sparrow search algorithm to obtain the improved sparrow search algorithm includes: In the process of foraging, the sparrow population is divided into three roles: explorers, followers and early warning; Assuming there are N sparrows in the group, the sparrow population is expressed as: In the formula, d represents the dimension of the sparrow population, n represents the number of sparrow populations, and x n,d represents the sparrow of the nth row and dth dimension; The fitness function of the sparrow population is calculated according to the sparrow population, and the fitness function expression is as follows: In the formula, f(x n,1 ,x n,2 …x n,d ) represents the fitness value of each sparrow in different dimensions; The explorer's position update formula is as follows: In the formula, t represents the current iteration number, T max Indicates setting the maximum number of iterations. represents the fitness value of the qth sparrow in the d-dimensional environment when the explorer has t iterations, α∈(0,1], R2(0,1], Q represents a random number that obeys a normal distribution, L is a matrix whose elements are all 1, ST is the warning threshold, R2<ST, indicating that no predator is found and foraging activities can be performed; R2≥ST, indicating that an invader appears and the sparrow population flies to other foraging areas; The follower's position update formula is as follows: In the formula, represents the best position record in the sparrow population, X worst Indicates the worst position record, A + is a matrix whose elements are all randomly assigned values ​​of 1 or -1. It represents the fitness value of the qth sparrow in the d-dimensional environment when the number of iterations of the follower is t; The position update formula of the alerter is as follows: In the formula, is the global optimal position record of the sparrow population, represents the global worst position record of the sparrow population, β is a random number that follows a normal distribution, K∈[-1,1], f q is the current individual fitness value, f g Indicates the fitness value of the best position in the current area, f w Represents the fitness value of the worst position in the current area, ε is a natural constant, It represents the fitness value of the qth sparrow in the d-dimensional environment when the number of iterations of the early warning is t.

5. The vehicle speed tracking control method based on the dynamic parameter adjustment neural network FOPID according to claim 4 is characterized in that: The step of improving the sparrow search algorithm to obtain the improved sparrow search algorithm includes: The improved explorer position update formula is as follows: In the formula, a1, a2 represent the individual learning parameters of the explorer, and c represents the dynamic parameter.

6. The vehicle speed tracking control method based on the dynamic parameter adjustment neural network FOPID according to claim 5 is characterized in that: The step of optimizing the control parameters of the FOPID controller by using the improved sparrow search algorithm to obtain the optimized control parameters of the FOPID controller includes: The improved fitness function of the sparrow population algorithm is expressed as follows: In the formula, b1, b2, b3, and b4 represent the weights of each part of the evaluation, e(t) is the system error, and t s Adjust the time for the system.

7. The vehicle speed tracking control method based on the dynamic parameter adjustment neural network FOPID according to claim 6 is characterized in that: The step of updating the control parameters of the optimized FOPID controller by using a neural network model to obtain the updated control parameters includes: The neural network model is associated with the FOPID controller to obtain a neural network-based FOPID controller model, so as to update the control parameters of the optimized FOPID controller; wherein the expression of the neural network-based FOPID controller model is as follows: In the formula, referring to Formula 4, α represents the fractional order, corresponding to λ in the FOPID controller, and the gradient in the neural network Corresponding to the system error e(t) in the FOPID controller, η represents the learning rate, corresponding to K in the FOPID controller p , δ corresponds to K in the FOPID controller i , x t Corresponding to u(t) in the FOPID controller.

8. The vehicle speed tracking control method based on the dynamic parameter adjustment neural network FOPID according to claim 7 is characterized in that: The step of updating the control parameters of the optimized FOPID controller by using a neural network model to obtain the updated control parameters includes: set up The neural network weight and bias update formulas for the neural network-based FOPID controller model are as follows: In the formula, t is the current iteration number, t+1 represents the next iteration, represents the neural network weight at iteration t, Represents the bias of the neural network at iteration t, l+1 represents the l+1th layer in the neural network, η represents the learning rate, and α is the fractional order.

9. The vehicle speed tracking control method based on the dynamic parameter adjustment neural network FOPID according to claim 8 is characterized in that: The step of updating the control parameters of the optimized FOPID controller by using a neural network model to obtain the updated control parameters includes: According to formula 13, the neural network weight and bias update formula of the neural network-based FOPID controller model is as follows:

10. The vehicle speed tracking control method based on the dynamic parameter adjustment neural network FOPID according to claim 9 is characterized in that: The step of controlling the vehicle model by the updated control parameters to achieve tracking control of the vehicle speed includes: A control output is adjusted according to the updated control parameter, and the control output is input into a vehicle model to control the vehicle speed.

Images