Adaptive Fuzzy Sliding Mode Guidance Control Method and System for High-Speed ​​Maglev Trains Based on Multi-Parameter Cooperative Optimization

Abstract

This invention discloses a multi-parameter collaborative optimization adaptive fuzzy sliding mode guidance control method and system for high-speed maglev trains. During train operation, a nonlinear state-space model of the guidance system considering strong crosswinds and passing disturbances is constructed. A basic sliding mode controller is designed based on this model. The basic sliding mode controller is applied to perform online estimation and feedforward compensation of the unknown total aerodynamic disturbance force, resulting in an adaptive fuzzy sliding mode controller. A fitness function based on a sparrow search algorithm is constructed to globally optimize the multi-dimensional parameters of the adaptive fuzzy sliding mode controller. This effectively suppresses the high-frequency voltage chattering phenomenon of traditional sliding mode control. It significantly improves the lateral disturbance resistance and operational stability of high-speed maglev trains under complex and strong aerodynamic disturbances. It also solves the bottleneck of difficult parameter tuning and susceptibility to local optima in traditional composite controllers.

Description

Technical Field

[0001] This invention belongs to the field of vehicle control, specifically relating to an adaptive fuzzy sliding mode guidance control method and system for high-speed maglev trains with multi-parameter collaborative optimization. Background Technology

[0002] Electromagnetic levitation (EMS) trains represent a significant development direction for achieving high-speed rail transit at 600 km / h. When operating at high speeds in the open, these trains are susceptible to sudden aerodynamic disturbances such as strong crosswinds and pressure waves from passing vehicles, leading to lateral shift of the car body (see attached stress diagram). Figure 2 This seriously affects operational stability.

[0003] To maintain lateral stability, the train primarily relies on the electromagnetic differential drive structure of the active guidance system (see attached reference for drive structure). Figure 1 ).

[0004] Because guide electromagnets inherently possess strong nonlinearity and open-loop instability, existing control methods and parameter tuning techniques still suffer from the following significant drawbacks when facing complex crosswind impacts:

[0005] Traditional sliding mode control faces the contradiction of being unable to suppress large chattering under strong disturbance rejection.

[0006] To improve the robustness of the guidance system to external disturbances, existing technologies often incorporate sliding mode control (SMC) algorithms. However, traditional sliding mode control inevitably relies on excessively large robust switching gain to suppress overall disturbances such as strong crosswinds. This approach leads to severe high-frequency chattering in the controller output voltage and can also negatively impact the dynamic stability of the guidance system, making the train highly susceptible to exceeding the safe air gap boundary under strong crosswinds.

[0007] Existing composite controllers have low multi-dimensional parameter tuning efficiency and are prone to getting trapped in local optima.

[0008] To alleviate the chattering problem in sliding mode control, some studies have attempted to introduce mechanisms such as fuzzy logic to construct composite controllers. However, the final performance of composite control is highly dependent on the selection of multidimensional control parameters.

[0009] Currently, automatic optimization of maglev control parameters often relies on manual trial and error or traditional intelligent algorithms such as particle swarm optimization (PSO) and genetic algorithms (GA). Manual trial and error struggles to obtain the globally optimal configuration when facing sudden strong winds; while traditional optimization algorithms generally suffer from drawbacks such as getting trapped in local optima and slow convergence when dealing with the high-dimensional parameter space of the guidance system control, failing to provide the system with an optimal parameter set that balances high wind resistance and low control chattering in real time. Summary of the Invention

[0010] The technical problem to be solved by the present invention is to provide a multi-parameter collaborative optimization adaptive fuzzy sliding mode guidance control method and system for high-speed maglev trains, which solves the problems of lateral deviation and high-frequency jitter of control voltage that high-speed maglev trains are prone to under external disturbances such as strong crosswinds in the prior art.

[0011] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0012] An adaptive fuzzy sliding mode guidance control method for high-speed maglev trains based on multi-parameter collaborative optimization includes the following steps:

[0013] Step 1: When the train is running, construct a nonlinear state-space model of the guidance system that takes into account strong crosswinds and passing disturbances.

[0014] Step 2: Design a basic sliding mode controller based on the nonlinear state-space model of the guidance system;

[0015] Step 3: Construct an adaptive fuzzy approximator based on the basic sliding mode controller to perform online estimation and feedforward compensation of the comprehensive disturbance caused by the total aerodynamic disturbance force, and obtain the adaptive fuzzy sliding mode controller.

[0016] Step 4: Construct the fitness function of the Sparrow Search Algorithm (SSA) to globally optimize the multidimensional parameters of the adaptive fuzzy sliding mode controller, and output the optimized adaptive fuzzy sliding mode guided controller.

[0017] In step 1, strong crosswinds and passing disturbances include the lateral crosswind interference force when a train is running in a crosswind environment and the lateral passing disturbance force when two trains meet. Among them, the lateral crosswind interference force... The following formula is used for calculation:

[0018]

[0019] in, air density, The effective frontal projection area of ​​the train from the side. The combined relative airflow velocity of the train and the crosswind. The aerodynamic coefficient for lateral force;

[0020] Lateral vehicle encounter disturbance force The following formula is used for calculation:

[0021]

[0022] in, For train speed, This is the transient lateral force coefficient during vehicle encounters.

[0023] The control law of the sliding mode controller is expressed by the following formula:

[0024]

[0025] In the formula, The differential terminal voltage output of the controller is used as the control input; For robust gain switching, For fuzzy basis functions, For sliding surface functions; and Design parameters for the sliding surface that are strictly greater than zero; , For nonlinear dynamic functions within the system, For the control gain function, This is the lateral offset. The lateral offset velocity, For differential control current, This is the online estimation vector for fuzzy weights.

[0026] The specific process of step 3 is as follows:

[0027] First, extract the air gap tracking error to construct the sliding surface. Based on this, the traditional sliding mode control law is calculated to ensure global asymptotic convergence in a perturbation-free state; simultaneously, the sliding surface is used... As the driving signal, it is adjusted in conjunction with the adaptive update law of weights. An adaptive fuzzy approximator is used to approximate the combined disturbance caused by the total aerodynamic disturbance force online, and the feedforward compensation is output in real time and synthesized into a differential control voltage. As the output of the adaptive fuzzy sliding mode controller.

[0028] In step 4, four core parameters are extracted from the adaptive fuzzy sliding mode control law to form the vector to be optimized, and global optimization of multi-dimensional parameters is performed. The four core parameters are the proportional sliding mode coefficient, the differential sliding mode coefficient, the robust switching gain, and the adaptive learning rate.

[0029] The fitness function is constructed by multiplying time by the integral of the absolute error, as follows:

[0030]

[0031] In the formula, To guide air gap error; For simulation time; This represents the total simulation duration.

[0032] A simulation system for an adaptive fuzzy sliding mode guidance control method for high-speed maglev trains based on multi-parameter collaborative optimization is provided. This system includes a MATLAB / Simulink interactive simulation platform. The sparrow search algorithm in the MATLAB script is responsible for initializing the population size and the maximum number of iterations, and generating candidate parameter solutions for each generation, which contain four core parameters to be optimized. The generated parameter sets are written to the basic working area in real time through assignment commands, and the background calls the Simulink guidance control physical model for time-domain simulation. The simulation model runs under a set first-order gust filter, calculates the guidance air gap error under crosswind disturbances of different intensities, and quantifies the fitness value of the current parameter individual based on the ITAE evaluation index.

[0033] The high-speed maglev train adaptive fuzzy sliding mode guidance control system with multi-parameter collaborative optimization includes an on-board control system, a car body, a bogie, left-side guide electromagnets and right-side guide electromagnets installed on the left and right sides of the bogie, and a ferromagnetic guide rail fixed on the ground. When the car body encounters crosswinds or sudden gas disturbances during oncoming traffic, the on-board control system uses the method described above to output a corresponding control voltage to the guide electromagnets, pulling the car body back to the center equilibrium position.

[0034] An electronic device includes a processor and a memory, wherein a computer program is stored in the memory, and the computer program, when executed by the processor, implements the method.

[0035] A computer-readable storage medium storing computer-readable instructions that, when executed by a processor, invoke the steps of the method.

[0036] Compared with the prior art, the present invention has the following beneficial effects:

[0037] 1. An adaptive fuzzy feedforward compensation mechanism is introduced, which effectively suppresses the high-frequency voltage chattering phenomenon of traditional sliding mode control.

[0038] 2. Significantly improved the lateral anti-disturbance capability and operational stability of high-speed maglev trains under complex and strong aerodynamic disturbances.

[0039] 3. An efficient multi-dimensional collaborative global optimization mechanism for control parameters is proposed, which solves the bottleneck of traditional composite controllers being difficult to tune parameters and prone to getting trapped in local optima. Attached Figure Description

[0040] Figure 1 The electromagnetic differential drive model is used for the guiding unit.

[0041] Figure 2 This is a lateral force model for the guidance system.

[0042] Figure 3This is a block diagram of the adaptive fuzzy sliding mode guidance control system of the present invention.

[0043] Figure 4 This is a flowchart of the sparrow search algorithm optimization process of the present invention.

[0044] Figure 5 This is a flowchart of the SSA-AFSMC joint optimization interaction based on MATLAB / Simulink in this invention.

[0045] Figure 6 This is the fitness iteration convergence curve of the SSA algorithm of this invention.

[0046] Figure 7 This is a comparison diagram of the guide air gap under a 3000N disturbance according to the present invention.

[0047] Figure 8 This is a comparison diagram of the convergence state of the sliding surface of the system under a 2000N disturbance according to the present invention.

[0048] Figure 9 This is a comparison diagram of the control voltage output waveform under a 3000N disturbance according to the present invention. Detailed Implementation

[0049] The structure and working process of the present invention will be further described below with reference to the accompanying drawings.

[0050] An adaptive fuzzy sliding mode guidance control method for high-speed maglev trains based on multi-parameter collaborative optimization includes the following steps:

[0051] Step 1: When the train is running, construct a nonlinear state-space model of the guidance system that takes into account strong crosswinds and passing disturbances.

[0052] Step 2: Design a basic sliding mode controller based on the nonlinear state-space model of the guidance system;

[0053] Step 3: Apply the basic sliding mode controller to perform online estimation and feedforward compensation of the unknown total aerodynamic disturbance force to obtain the adaptive fuzzy sliding mode controller;

[0054] Step 4: Construct the fitness function of the sparrow search algorithm to globally optimize the multidimensional parameters of the adaptive fuzzy sliding mode controller.

[0055] Specific embodiments, such as Figures 1 to 9 As shown:

[0056] The overall technical process is further elaborated below through steps S10 to S50;

[0057] S30 and S40 both belong to overall step 3, which includes the construction process of the adaptive fuzzy approximator and the synthesis process of the adaptive fuzzy sliding mode control law, that is, first establish the adaptive fuzzy approximator, and then synthesize the adaptive fuzzy sliding mode controller.

[0058] The adaptive fuzzy sliding mode guidance control method for high-speed maglev trains with multi-parameter collaborative optimization includes the following steps:

[0059] Step S10: During train operation, a nonlinear state-space model of the guidance system considering strong crosswinds and passing disturbances is constructed. Specifically, the guidance system of the conventional electromagnetic levitation train adopts a differential drive structure of electromagnets on both sides, with a guide air gap of 10mm on both sides during normal operation. Ignoring the edge effect and leakage flux of the air gap magnetic field, and assuming that the permeability of the core and guide rail tends to infinity, the electromagnetic attraction force of the electromagnet on one side... The expression is:

[0060] (1)

[0061] in, The permeability of free space, The effective cross-sectional area of ​​the electromagnet poles. The number of coil turns. For coil current, This refers to the length of the single-sided guide air gap.

[0062] The nonlinear voltage balance equation for a single-sided electromagnet is:

[0063] (2)

[0064] In the formula, This is the equivalent DC resistance of the electromagnet coil; Coil current The first derivative with respect to time; The first derivative of the length of the single-sided guide air gap with respect to time; the dot above the symbol indicates the first derivative with respect to time; This is the voltage across the coil of a single electromagnet.

[0065] Appendix Figure 1 middle, The voltage at the coil terminals of the left-side guiding electromagnet. This is the voltage at the coil end of the right-side guide electromagnet.

[0066] Let the balance air gap be The balance current is The train deviated laterally after being disturbed. The control system applies differential current. The air gaps and currents on the left and right sides are as follows:

[0067] (3)

[0068] In the formula, and These are the left guide air gap and the right guide air gap, respectively. and These are the currents of the left and right electromagnet coils, respectively. The rated guide air gap on one side when the train is in the center equilibrium position; To balance the current; This represents the lateral offset of the vehicle body relative to its central equilibrium position. This is for differential control current.

[0069] In addition, this embodiment specifies that the train's lateral leftward deviation is the positive direction, and the signs of all subsequent electromagnetic forces, aerodynamic disturbance forces, and lateral displacements are determined according to this positive direction.

[0070] Net guiding force acting on the train The difference between the electromagnetic forces on the left and right sides:

[0071] (4)

[0072] In the formula, The electromagnetic attraction generated by the left-side guide electromagnet; The electromagnetic attraction generated by the right-side guide electromagnet; It is the net guiding force formed by the electromagnetic attraction on both the left and right sides.

[0073] Combined relative airflow velocity of the train and the crosswind Its sideslip angle β can be expressed as:

[0074]

[0075] In the formula, The train's operating speed; The natural crosswind speed varies over time; The combined relative airflow velocity formed by the train's running speed and the natural crosswind speed; The sideslip angle for the composite relative airflow; For time.

[0076] According to aerodynamic principles, the lateral crosswind interference force acting on the area where a single guide frame is located... for:

[0077] (5)

[0078] in, air density, The effective frontal projection area of ​​the train from the side. The combined relative airflow velocity of the train and the crosswind. This represents the aerodynamic coefficient for lateral force. It also represents the lateral disturbance force during the meeting of two trains. for:

[0079] (6)

[0080] When two high-speed maglev trains meet in opposite directions at the same speed, their relative speeds are approximately: Therefore, adopt Calculate the transient aerodynamic effects generated during the passing process. Among them, For train speed, Here is the transient lateral force coefficient during oncoming traffic. The total aerodynamic disturbance force on the system, including crosswind and oncoming traffic disturbances, is:

[0081] (7)

[0082] In the formula, The total aerodynamic disturbance force acting on the guide unit in the lateral direction is composed of lateral crosswind disturbance force. Disturbance force when meeting oncoming traffic The result is obtained by superposition.

[0083] Select the lateral offset of the vehicle body Lateral offset speed Differential control current As a state variable, the differential terminal voltage To control the input, the nonlinear state-space equations of the guidance system are rearranged into the following affine nonlinear standard form:

[0084] (8)

[0085] In the formula, The lateral equivalent mass of the guiding unit; This represents the lateral acceleration generated by the total aerodynamic disturbance force.

[0086] Among them, the nonlinear dynamic function inside the system , and control gain function They are respectively:

[0087] (9)

[0088] The function of transverse nonlinear acceleration generated by the differential electromagnetic attraction of the left and right guide electromagnets; This is the internal nonlinear dynamic function of the differential control current; The control gain function of differential control voltage on differential control current;

[0089] The above, For the equivalent quality of the guiding unit, This is the equivalent inductance related to the air gap.

[0090] in, To determine the lateral offset of the vehicle body The calculated equivalent inductance is related to the actual guide air gap; when modeling using the left-side guide air gap, When using right-side guide air gap modeling,

[0091] Step S20: Design a basic sliding mode control strategy based on the state-space model of the guidance system; specifically, adjust the differential terminal control voltage. This causes the lateral offset to It can quickly and without static error track the rated guide air gap. During normal centering operation:

[0092] Define the system's position tracking error and its derivatives as follows:

[0093] (10)

[0094] This refers to the lateral position tracking error. This is the first time derivative of the lateral position tracking error; This is the second time derivative of the lateral position tracking error; This represents the desired lateral offset.

[0095] To ensure asymptotic convergence of the error state, a third-order linear sliding surface function is selected as shown below:

[0096] (11)

[0097] In the formula, For sliding surface functions; The proportional sliding coefficient determines the degree of influence of lateral position error on the sliding surface; The differential sliding coefficients are strictly positive sliding surface design parameters that must satisfy the Hurwitz polynomial condition to ensure the system operates on the sliding mode surface. Gradual stabilization.

[0098] By taking the time derivative of the sliding surface function and combining it with the nonlinear state-space equations of the guiding system, the basic sliding mode control law is obtained as follows:

[0099] (12)

[0100] In the formula, It is the differential control voltage output by the basic sliding mode controller; To robustly switch gain. Nonlinear dynamic function For state variables The partial derivatives;

[0101] Nonlinear dynamic function For state variables The partial derivatives;

[0102] It is a symbolic function;

[0103] This is a robust switching gain used to compensate for residual errors in fuzzy approximation.

[0104] The above basic sliding mode control law uses a robust switching term. Suppress external disturbances. When the total aerodynamic disturbance force is large, the robust switching gain needs to be increased accordingly. This can easily cause high-frequency fluctuations in the differential control voltage. Therefore, in step S30, an adaptive fuzzy approximator is constructed to estimate the unknown comprehensive disturbance online, and in step S40, the disturbance estimate is introduced into the basic sliding mode control law.

[0105] Step S30: Construct an adaptive fuzzy approximator to perform online estimation and feedforward compensation of the unknown total aerodynamic disturbance force; specifically, using a sliding mode surface function... As input to the fuzzy approximator, the sliding mode approximates the comprehensive perturbation estimate in the dynamics. As output, a fuzzy IF-THEN rule is designed, employing a single-valued fuzzy logic generator, product inference, and center-average defuzzification method. A fuzzy basis function is introduced. As shown below:

[0106] (13)

[0107] In the formula, This is the sequence number of the current fuzzy rule; The sequence number of the fuzzy rule during the summation process; For the first The input fuzzy set corresponding to each rule; For sliding surface functions For fuzzy sets Membership degree; The total number of fuzzy rules; For the first A normalized fuzzy basis function.

[0108] In this embodiment, based on the sliding surface function Preset input range during the operation of the guidance system ,set up Each Gaussian membership function has its center evenly distributed within the preset input range to ensure that the fuzzy rules can cover the entire input range of the sliding surface function.

[0109] No. A Gaussian membership function is represented as follows:

[0110] The center of each membership function Set it according to the following formula:

[0111]

[0112] The interval between adjacent centers is:

[0113]

[0114] The width of each Gaussian membership function is set as follows:

[0115] In the above formula, and These are the lower and upper limits of the input range for the sliding surface function, respectively. The total number of fuzzy rules; For fuzzy rule numbers; For the first The center of a Gaussian membership function; For the first The width of a Gaussian membership function; The interval between the centers of adjacent Gaussian membership functions; For sliding surface functions For the A set of input fuzzy sets The degree of membership.

[0116] In one specific implementation, the total number of fuzzy rules Set to 7; Input range for sliding surface function The sliding surface function is determined based on its maximum allowable variation range during the simulation of the guiding system. When the simulation pre-run yields the sliding surface function primarily distributed in... (In the formula, This is a preset upper bound for the absolute value of the sliding surface function obtained through pre-simulation. Within this bound, take... , And set the centers of each Gaussian membership function according to the above equal spacing method.

[0117] Ideal fuzzy weight vector, Online estimation vector of ideal fuzzy weights : Fuzzy weight estimation error vector. Define the corresponding fuzzy basis function vector. Then the optimal approximation of the fuzzy system for unknown aerodynamic disturbances can be transformed into the following linearly parameterized inner product form.

[0118] (14)

[0119] In the formula, This represents the estimate of the unknown disturbance by the fuzzy system. This is the transpose of the ideal weight vector; For fuzzy basis function vectors; Ideal fuzzy weight vector. According to the universal approximation theorem, a fuzzy system can approximate the total unknown nonlinear disturbance with arbitrary precision. And there exists a minimum approximation error. .

[0120] Step S40: The comprehensive disturbance estimate obtained in step S30 is introduced into the basic sliding mode control law established in step S20 to obtain the adaptive fuzzy sliding mode overall control law.

[0121]

[0122] In the formula, The differential control voltage output by the adaptive fuzzy sliding mode controller; This is the online estimation vector for fuzzy weights; For fuzzy basis function vectors; This is the comprehensive perturbation estimate output by the adaptive fuzzy approximator; the meanings of the other symbols are the same as those defined in step S20.

[0123] To ensure the global stability of the guidance system under complex perturbations and to guarantee the convergence of the fuzzy parameter adaptive learning process, based on Lyapunov stability theory, a method considering the sliding surface state is selected. With parameter estimation error Positive definite composite functional .

[0124] In the formula, For composite Lyapunov functions; This represents the error vector for fuzzy weight estimation. The adaptive learning rate is greater than zero.

[0125] To eliminate the impact of unknown parameter estimation errors on system stability, an adaptive update law for fuzzy weights is designed as follows:

[0126] (15)

[0127] In the formula, The first-order time derivative of the online estimate of the fuzzy weight vector; For adaptive learning rate, For sliding surface functions; Let be the fuzzy basis function vector. This update law ensures that the synthesized Lyapunov function of the system is non-increasing and has a lower bound. Under the conditions that the total aerodynamic disturbance is bounded, the fuzzy approximation error is bounded, and the robust switching gain satisfies the stability condition, it guarantees that the signals inside the closed-loop system are bounded and makes the lateral position tracking error asymptotically converge.

[0128] Step S50: Construct the fitness function of the sparrow search algorithm to globally optimize the multidimensional parameters of the adaptive fuzzy sliding mode controller; specifically, extract the four core parameters from the adaptive fuzzy sliding mode control law to form the vector to be optimized. (i.e., proportional sliding mode coefficient) Differential sliding mode coefficient Robust switching gain and adaptive learning rate Here, is a candidate control parameter vector, not representing the state variables of the steering system. To quantify the quality of the parameters, the fitness function is constructed using the time-integral-absolute-error (ITAE) method, as shown below:

[0129] (16)

[0130] In the formula, This refers to the lateral position tracking error. For simulation time; The total duration of a single simulation is denoted as . An initial candidate solution population is randomly generated within the search boundary and divided into three roles: discoverer, joiner, and scout. These roles collaboratively perform wide-area global search of the parameter space, deep local exploitation, and random reverse perturbation to prevent getting trapped in local extrema. Each set of candidate parameters is substituted into the guiding physics model to calculate the fitness value. Based on the fitness distribution of the population, the three roles are driven to iteratively update their spatial positions. Boundary checks are performed on all individuals. If a lower error is generated, the globally optimal parameter set is updated until the maximum number of iterations is reached. Finally, the optimal candidate parameter combination that minimizes the ITAE fitness value of the lateral position tracking error is output. The improvement effect of this parameter combination on high-frequency chattering is evaluated through subsequent control voltage comparison simulations.

[0131] In this embodiment, to ensure the feasibility of the sparrow search algorithm's optimization process, the population size of the sparrow search algorithm is set. The maximum number of iterations is 30. It is 100. Since this embodiment only considers the proportional sliding mode coefficient... Differential sliding mode coefficient Robust switching gain and adaptive learning rate The four control parameters are jointly optimized, and 30 candidate individuals can balance the global search capability with the computational efficiency of MATLAB / Simulink co-simulation; 100 iterations can ensure that the fitness function fully converges.

[0132] In terms of population role division, individuals in the population are sorted according to their fitness values. The top 20% of individuals with the best fitness are selected as discoverers to guide the population in a wide-area global search within the parameter space. The remaining individuals are designated as joiners to explore locally around the currently optimal parameter region. To reduce the risk of the algorithm getting stuck in local optima, in each iteration, 10%–20% of the individuals in the entire population are randomly selected as scouts, and these scouts are updated with random perturbations to enhance the algorithm's ability to escape local optima.

[0133] In this embodiment, the search boundaries for the four control parameters to be optimized are set as shown in the table below:

[0134]

[0135] The search boundary is determined based on the acceptable range of the lateral offset response of the guidance system, the amplitude of the control voltage output, and the adaptive fuzzy weight update rate. Specifically, the proportional sliding mode coefficient mainly affects the convergence rate of the lateral position error; the differential sliding mode coefficient mainly affects the damping characteristics of the lateral offset rate; the robust switching gain is used to compensate for the residual error in fuzzy approximation and unmodeled disturbances; and the adaptive learning rate is used to adjust the online update rate of the fuzzy weights.

[0136] like Figure 5 As shown, in the specific optimization process, the search boundary is first randomly generated. There are initial candidate parameter vectors, each of which is represented as . Subsequently, each set of candidate parameters is written into the MATLAB basic workspace, and a Simulink guided system model is called for time-domain simulation to calculate the ITAE fitness value corresponding to that set of parameters. The SSA algorithm updates the positions of the discoverer, joiner, and scout based on the fitness distribution of all individual candidate parameters, and performs boundary checks after each update; when a candidate parameter exceeds the preset search boundary, it is restricted back to the boundary range of the corresponding parameter. If the updated candidate parameter vector obtains a smaller ITAE fitness value, the current globally optimal parameter vector is updated. This process is repeated until the maximum number of iterations is reached. The final output is the optimal candidate parameter combination that minimizes the lateral position tracking error (ITAE) fitness value. .

[0137] Based on the above method, this invention also provides an adaptive fuzzy sliding mode (AFSMC) guidance control system optimized by the Sparrow Search Algorithm (SSA), applied to a conventional electromagnetic attraction-type high-speed maglev train. The system's composition, working principle, and operation process are as follows:

[0138] 1. Description of physical structure and working principle (in conjunction with appendix) Figure 1 Appendix Figure 2 )

[0139] Regarding the hardware physical structure upon which this invention is based, in conjunction with the appendix... Figure 1 (Electromagnetic differential drive model of the guide unit) and appendix Figure 2 (Lateral Force Model of the Guiding System) This active guidance system mainly includes: the car body, the bogie, the left and right guide electromagnets installed on the left and right sides of the bogie, and the ferromagnetic guide rail fixed to the ground. Its basic working principle is as follows: Under normal operating conditions, a rated air gap of 10mm is maintained between the pole faces of the left and right guide electromagnets and the sides of the ferromagnetic guide rail. When the train is suddenly disturbed to the left by an external crosswind, the left air gap increases and the right air gap decreases; at this time, the onboard control system adopts a differential drive strategy, increasing the control voltage on the left side. And reduce the control voltage on the right side. This causes the electromagnet on the left to generate an electromagnetic attraction. Greater than the electromagnetic attraction force generated by the electromagnet on the right. This generates a net guiding force to the right, pulling the vehicle back to its central equilibrium position.

[0140] 2. Control method workflow description (in conjunction with appendix) Figure 3 Appendix Figure 4 )

[0141] To achieve efficient and stable control of the aforementioned hardware architecture, the control method in this embodiment specifically includes the following steps:

[0142] Step S10: Construct a nonlinear state-space model of the guidance system that considers strong crosswinds and oncoming vehicle disturbances;

[0143] Specifically, the total external nonlinear aerodynamic disturbance experienced by the train during open-air operation includes natural crosswind interference force and transient pulse force from passing vehicles. A lateral offset is selected. Lateral offset speed and differential control current As the state variables of the system, a single-input single-output affine nonlinear standard state-space equation is constructed by combining Newton's second law and the nonlinear voltage balance equation of an electromagnet for controller design.

[0144] The purpose of this step is to solve the problem of unclear description of the stress state of the car body under complex flow field environment in the background technology, clarify the physical mechanism of crosswind disturbance on the lateral stability of the train from the dynamics level, and provide accurate mathematical model support for the subsequent controller design.

[0145] Step S20: Design a basic sliding mode control strategy based on the state-space model of the guidance system;

[0146] Specifically, in conjunction with the appendix Figure 3 (Block diagram of adaptive fuzzy sliding mode guidance control system), defining the system's position tracking error. Select a third-order linear sliding surface function Introducing total crosswind disturbance The lateral offset of the vehicle body is obtained by modeling the guidance system.

[0147] Step S30: Construct an adaptive fuzzy approximator to perform online estimation and feedforward compensation of the unknown total aerodynamic disturbance force;

[0148] Specifically, because crosswinds and oncoming traffic disturbances in actual operation are highly nonlinear and difficult to measure accurately, this embodiment introduces a fuzzy logic system. For example... Figure 3 The compensation loop shown selects a sliding surface function. As input to the fuzzy approximator, a fuzzy IF-THEN rule is designed. Using this fuzzy rule, the optimal approximation of the unknown total aerodynamic disturbance force is transformed into a linearly parameterized inner product form. .

[0149] Step S40: Design an adaptive fuzzy sliding mode overall control law in conjunction with a fuzzy approximator;

[0150] Specifically, the feedforward compensation term from the fuzzy system output in step S30 is embedded into the sliding mode control architecture in step S20 to synthesize an adaptive fuzzy sliding mode overall control law. To ensure the stability of the closed-loop system, an adaptive update law for fuzzy weights is designed based on Lyapunov theory. .

[0151] Step S50: Construct the fitness function of the Sparrow Search Algorithm (SSA) to globally optimize the multidimensional parameters of the controller;

[0152] Specifically, in conjunction with the appendix Figure 4 (Flowchart of the Sparrow Search Algorithm Optimization) Extracting four core parameters from the adaptive fuzzy sliding mode control law: proportional sliding mode coefficient Differential sliding mode coefficient Robust switching gain and adaptive learning rate A four-dimensional parameter vector to be optimized is constructed. The fitness function is built using the Time-Integral-Absolute-Error (ITAE). The population is divided into three roles: discoverers, joiners, and scouts. Fitness is calculated and parameter positions are updated iteratively in a co-simulation environment until the globally optimal parameter set that minimizes the air gap error is output. In this embodiment, a MATLAB / Simulink simulation model of the guidance system is established, with a rated single-sided guide air gap of 10mm. SSA parameter optimization is performed under 2000N side wind disturbance dynamics. The disturbance rejection performance of the controller is verified using three-stage step side wind disturbance dynamics of 1000N, 2000N, and 3000N. The comparison objects are the AFSMC without SSA optimization and the SSA-AFSMC with SSA optimization.

[0153] The final optimized parameters are shown in the table below.

[0154]

[0155] 3. Simulation verification and positive effects of the embodiments (in conjunction with the appendix) Figure 6 To be continued Figure 9 )

[0156] To verify the effectiveness of the present invention, in conjunction with the appendix... Figure 6 To be continued Figure 9 The simulation results will be further explained below:

[0157] (1) As attached Figure 6 As shown in the fitness iteration convergence curve of the SSA algorithm, the algorithm exhibits a multi-level step-down characteristic during the optimization process, effectively avoiding local optima, and achieving global optimal convergence around the 70th generation.

[0158] (2) As attached Figure 7 (Comparison of guide air gap under 3000N disturbance) and attached Figure 8 (Comparison of system sliding surface convergence state under 2000N disturbance) As shown in the figure, under step crosswind impacts of 1000N to 3000N, compared with the unoptimized traditional method, the optimized system (SSA-AFSMC) of this invention exhibits a smaller sliding surface displacement and can reconstruct and converge extremely quickly. Under extreme crosswinds of 3000N, this invention can strictly limit the maximum dynamic displacement of the guide air gap to within 0.3mm, greatly improving the running stability of the train when disturbed.

[0159] (3) As attached Figure 9 As shown in the comparison diagram of control voltage output waveform under 3000N disturbance, the optimized strategy of this invention reduces the peak-to-peak value of the control voltage by nearly 36%. This proves that the present invention effectively suppresses the high-frequency voltage chattering phenomenon inherent in sliding mode control, reduces the fluctuation amplitude of the control voltage, and helps to reduce the frequent adjustment of the guide electromagnet drive circuit.

[0160] An electronic device includes a processor and a memory, wherein a computer program is stored in the memory, and the computer program, when executed by the processor, implements the method.

[0161] A computer-readable storage medium storing computer-readable instructions that, when executed by a processor, invoke the steps of the method.

[0162] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, essentially, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0163] Finally, it should be noted that the above specific embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications and substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A multi-parameter collaborative optimization adaptive fuzzy sliding mode guidance control method for high-speed maglev trains, characterized by: Includes the following steps: Step 1: When the train is running, construct a nonlinear state-space model of the guidance system that takes into account strong crosswinds and passing disturbances. Step 2: Design a basic sliding mode controller based on the nonlinear state-space model of the guidance system; Step 3: Construct an adaptive fuzzy approximator based on the basic sliding mode controller to perform online estimation and feedforward compensation of the comprehensive disturbance caused by the total aerodynamic disturbance force, and obtain the adaptive fuzzy sliding mode controller. Step 4: Construct the fitness function of the sparrow search algorithm to globally optimize the multidimensional parameters of the adaptive fuzzy sliding mode controller, and output the optimized adaptive fuzzy sliding mode guided controller.

2. The adaptive fuzzy sliding mode guidance control method for high-speed maglev trains with multi-parameter collaborative optimization according to claim 1, characterized in that: In step 1, strong crosswinds and passing disturbances include the lateral crosswind interference force when a train is running in a crosswind environment and the lateral passing disturbance force when two trains meet. Among them, the lateral crosswind interference force... The following formula is used for calculation: in, air density, The effective frontal projection area of ​​the train from the side. The combined relative airflow velocity of the train and the crosswind. The aerodynamic coefficient for lateral force; Lateral vehicle encounter disturbance force The following formula is used for calculation: in, For train speed, This is the transient lateral force coefficient during vehicle encounters.

3. The adaptive fuzzy sliding mode guidance control method for high-speed maglev trains with multi-parameter collaborative optimization according to claim 2, characterized in that: The control law of the sliding mode controller is expressed by the following formula: In the formula, The differential terminal voltage output of the controller is used as the control input; For robust gain switching, For fuzzy basis functions, For sliding surface functions; and Design parameters for the sliding surface that are strictly greater than zero; , For nonlinear dynamic functions within the system, For the control gain function, This is the lateral offset. The lateral offset velocity, For differential control current, This is the online estimation vector for fuzzy weights.

4. The adaptive fuzzy sliding mode guidance control method for high-speed maglev trains with multi-parameter collaborative optimization according to claim 3, characterized in that: The specific process of step 3 is as follows: First, extract the air gap tracking error to construct the sliding surface. Based on this, the traditional sliding mode control law is calculated to ensure global asymptotic convergence in a perturbation-free state; simultaneously, the sliding surface is used... As the driving signal, it is adjusted in conjunction with the adaptive update law of weights. An adaptive fuzzy approximator is used to approximate the combined disturbance caused by the total aerodynamic disturbance force online, and the feedforward compensation is output in real time and synthesized into a differential control voltage. As the output of the adaptive fuzzy sliding mode controller.

5. The adaptive fuzzy sliding mode guidance control method for high-speed maglev trains with multi-parameter collaborative optimization according to claim 4, characterized in that: In step 4, four core parameters are extracted from the adaptive fuzzy sliding mode control law to form the vector to be optimized, and global optimization of multi-dimensional parameters is performed. The four core parameters are the proportional sliding mode coefficient, the differential sliding mode coefficient, the robust switching gain, and the adaptive learning rate.

6. The adaptive fuzzy sliding mode guidance control method for high-speed maglev trains with multi-parameter collaborative optimization according to claim 5, characterized in that: The fitness function is constructed by multiplying time by the integral of the absolute error, as follows: In the formula, To guide air gap error; For simulation time; This represents the total simulation duration.

7. A simulation system for implementing the multi-parameter collaborative optimization adaptive fuzzy sliding mode guidance control method for high-speed maglev trains as described in any one of claims 1 to 6, characterized in that: The system includes a MATLAB / Simulink interactive simulation platform. The sparrow search algorithm in the MATLAB script is responsible for initializing the population size and the maximum number of iterations, and generating candidate parameter solutions for each generation containing four core parameters to be optimized. The generated parameter sets are written to the basic workspace in real time through assignment commands, and the background calls the Simulink steering control physical model to perform time-domain simulation. The simulation model runs under a set first-order gust filter, calculates the steering air gap error under crosswind disturbances of different intensities, and quantifies the fitness value of the current parameter individual based on the ITAE evaluation index.

8. A high-speed maglev train adaptive fuzzy sliding mode guidance control system with multi-parameter collaborative optimization, characterized in that: The system includes an onboard control system, a car body, a bogie, a left-side guide electromagnet and a right-side guide electromagnet installed on the left and right sides of the bogie, and a ferromagnetic guide rail fixed on the ground. When the car body encounters crosswinds or sudden gas disturbances during oncoming traffic, the onboard control system uses the method described in any one of claims 1 to 6 to output a corresponding control voltage to the guide electromagnets to pull the car body back to the center balance position.

9. An electronic device, characterized in that, It includes a processor and a memory, wherein a computer program is stored in the memory, and the computer program, when executed by the processor, implements the method as described in any one of claims 1 to 6.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-readable instructions that, when executed by a processor, invoke the steps of the method according to any one of claims 1 to 6.

Images