A method of controlling the stability of a vehicle with a blown tire

Abstract

The application discloses a tire burst vehicle stability control method, which adopts LS-DYNA to perform three-dimensional modeling simulation analysis on a tire, simulates the inflation and tire burst process by applying a load to the inner surface of the tire, obtains the nonlinear attenuation law of the radial stiffness, the longitudinal stiffness and the lateral stiffness of the tire after the tire burst compared with the normal tire, revises key parameters based on the UniTire tire model theory, establishes a tire burst tire model capable of representing the mechanical characteristics after the tire burst and integrates the tire burst tire model into a vehicle dynamics model, can accurately reproduce the nonlinear mechanical evolution law of the tire under the tire burst condition, provides an accurate dynamics basis for the design and verification of a subsequent stability control algorithm, and improves the working condition adaptation precision of the control scheme.

Description

Technical Field

[0001] This invention belongs to the field of vehicle engineering technology, and in particular relates to a method for controlling the stability of a vehicle with a tire blowout. Background Technology

[0002] Distributed drive electric vehicles adopt a four-wheel independent motor drive architecture, and the torque of each wheel can be independently, precisely and quickly adjusted. Compared with traditional vehicles that rely on braking to achieve stability control, they have higher flexibility and response speed in terms of driving stability control, providing a new technical path for the safety control of vehicles under extreme conditions.

[0003] Tire blowout is an extremely dangerous and sudden situation during driving. After a tire blows out, the radial stiffness and lateral stiffness of the failed tire decrease drastically, causing an asymmetric transfer of vehicle load and disorder in tire contact friction characteristics. This leads to sudden changes in vehicle yaw rate and loss of lateral control, seriously threatening driving safety. The core challenge of tire blowout lies in the instantaneous and strong abrupt change in vehicle dynamics. The system exhibits strong nonlinearity and time-varying parameter characteristics, placing extremely high demands on the adaptability and response speed of stability control algorithms.

[0004] Current vehicle stability control methods are mostly based on linear dynamic models designed under normal driving conditions, which are difficult to adapt to the drastic parameter changes and nonlinear characteristics of the system after a tire blowout. Under the distributed drive architecture, existing solutions fail to fully utilize the control advantages of four-wheel independent drive, lack multi-wheel torque coordination optimization methods for tire blowout conditions, and are unable to quickly suppress tire blowout disturbances and reconstruct the vehicle's stable operating boundary. They also lack a coordinated mechanism for steering correction and torque stability control, and cannot simultaneously take into account attitude stability and path maintenance. The control effect under tire blowout conditions is still significantly insufficient. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention proposes a method for controlling vehicle stability in the event of a tire blowout.

[0006] The technical solution of the present invention is as follows:

[0007] A tire blowout vehicle stability control method, executed after a tire blowout is detected, includes:

[0008] The vehicle’s current yaw rate and center of gravity sideslip angle are obtained and compared with the corresponding target reference values ​​obtained based on the vehicle dynamics model to obtain the deviation and rate of change of the two state variables.

[0009] The deviation and the rate of change of deviation are input into the fuzzy PID controller, and the additional yaw moment command for suppressing vehicle instability is generated by online adaptive adjustment of the PID gain parameter.

[0010] The dream optimization algorithm is used to solve the problem with the four-wheel torque distribution coefficient as the optimization variable to obtain the optimal torque distribution coefficient for four-wheel independent drive that can realize the additional yaw moment. At the same time, the pure tracking algorithm is used to select the target point at a preset aiming distance in front of the vehicle on the reference path and solve the front wheel steering angle command used to correct the vehicle's driving direction.

[0011] The output torque of the four wheels is controlled according to the optimal torque distribution coefficient, and the vehicle steering is controlled according to the front wheel steering angle command to achieve vehicle driving stability control after a tire blowout.

[0012] Furthermore, the target reference value is obtained in the following way: a state-space model is established based on the two-degree-of-freedom vehicle dynamics theory, the ideal reference state of the vehicle's lateral dynamics is derived under the steady-state driving assumption, and then the road surface adhesion coefficient is introduced as a physical constraint boundary to correct the limit values ​​of the yaw rate and the center of gravity sideslip angle, and the target expected state of the vehicle under the current working condition is calculated. The target expected state includes the target reference values ​​corresponding to the yaw rate and the center of gravity sideslip angle.

[0013] Furthermore, when the fuzzy PID controller is working, the deviation between the actual operating state of the vehicle and the target desired state, as well as the rate of change of the deviation, are used as the input of the fuzzy PID controller. The proportional, integral and derivative gain parameters are adaptively tuned online through preset fuzzy rules, and the output is used to counteract the additional yaw moment of the tire blowout disturbance.

[0014] Furthermore, the process of solving the optimal torque distribution coefficient using the dream optimization algorithm is divided into three stages: initialization, exploration, and development. The global optimization and local search capabilities are balanced through basic memory strategy, forgetting and supplementation strategy, and dream sharing strategy.

[0015] Furthermore, during the initialization phase, a random population is generated in the search space as an initial solution to initiate the optimization process; at the same time, the basic memory strategy is used to call the steady-state control experience before the tire blowout as part of the initial solution of the initial population to shorten the control response delay in the early stage of the tire blowout.

[0016] Furthermore, during the exploration phase, the population is divided into multiple groups based on differences in memory capacity. Before each iteration considered as dream behavior, individuals within a group are reset to their historical best position. Then, some dimensions are randomly selected as forgetting dimensions for position updates. Through forgetting and replenishment strategies, old parameter constraints that have failed due to tire blowouts are dynamically eliminated. The algorithm adaptively searches for the optimal torque combination that matches the new tire state. At the same time, an information sharing mechanism within the group is introduced to enhance the algorithm's ability to escape local optima.

[0017] Furthermore, during the development phase, all individuals within the population share the global optimal solution. In the non-forgetting dimension, the position information of the group's optimal solution is retained to quickly move towards the optimal solution. In the forgetting dimension, the position is updated through random perturbation and cosine function self-organization to achieve local fine search. At the same time, the torque distribution among the four wheels is coordinated through a dream sharing strategy.

[0018] Furthermore, to address the torque optimization allocation requirements of distributed drive electric vehicles under tire blowout conditions, the population size of the dream optimization algorithm is set to 8, the total number of iterations is set to 150, and a 4-dimensional particle dimension is selected to correspond to the independent torque allocation of the four wheels. The maximum number of iterations in the exploration phase is set to nine-tenths of the total maximum number of iterations, the number of forgotten dimensions is dynamically set according to the problem dimension, and differentiated boundary processing strategies are adopted for high- and low-dimensional problems.

[0019] Furthermore, when using a pure tracking algorithm to solve for the front wheel steering angle command, a point on the reference path at a preset aiming distance from the center point of the vehicle's rear axle is first selected as the tracking target. Based on the geometric relationship between the vehicle's current position, heading, and the target point, the correspondence between the turning radius, aiming distance, and yaw angle is determined using the sine theorem, and the trajectory curvature is calculated. Then, based on the Ackermann steering geometry principle and the vehicle's wheelbase, the relationship between the front wheel steering angle and curvature is established. Finally, the desired front wheel steering angle that allows the vehicle to smoothly follow the predetermined path is solved.

[0020] Furthermore, the method also includes an offline step of pre-constructing a blowout tire model: using LS-DYNA explicit dynamics software to establish a tire finite element model, simulating the tire inflation and load-bearing process by applying air pressure loads to the inner surface of the tire and contact constraints with the moving road surface, and simulating and analyzing the nonlinear decay law of the tire's longitudinal stiffness and lateral stiffness under the blowout condition, then constructing a vehicle dynamics model based on the UniTire tire model theory, and correcting the key parameters of the UniTire model according to the stiffness decay law, and integrating the blowout tire model characterizing the mechanical properties after the blowout into the vehicle dynamics model.

[0021] Compared with the prior art, the present invention has the following beneficial effects:

[0022] This invention provides a vehicle stability control method for tire blowout. The method uses LS-DYNA for three-dimensional modeling and simulation analysis of the tire. By applying loads to the inner surface of the tire, it simulates the inflation and blowout process, obtaining the nonlinear attenuation laws of radial stiffness, longitudinal stiffness, and lateral stiffness of the blowout tire compared to a normal tire. Then, based on the UniTire tire model theory, key parameters are corrected to establish a blowout tire model that characterizes the mechanical properties after a blowout, which is integrated into the vehicle dynamics model. This method can accurately reproduce the nonlinear mechanical evolution of the tire under blowout conditions, providing an accurate dynamic basis for the design and verification of subsequent stability control algorithms and improving the condition adaptability accuracy of the control scheme.

[0023] This invention addresses the problems of sudden changes in vehicle dynamic parameters and strong nonlinearity leading to poor adaptability of fixed-parameter controllers and difficulty in quickly suppressing instability trends after a tire blowout. It establishes a two-layer collaborative control architecture. The upper layer employs a fuzzy PID control strategy, using the deviation and rate of change of the vehicle's actual yaw rate and sideslip angle relative to reference values ​​as input. Through a preset fuzzy rule table, the proportional, integral, and derivative gains of the PID controller are adjusted online to generate the optimal additional yaw torque command in real time. This adaptively matches the changing operating conditions after a tire blowout, effectively improving the dynamic response speed and steady-state accuracy of the control. For distributed drive electric vehicles, the complex four-wheel torque distribution under tire blowout conditions and the tendency of traditional optimization algorithms to get trapped in local optima are addressed. The lower layer employs a dream-based optimization algorithm to optimize the torque distribution coefficient of the additional yaw torque. The algorithm operates through three stages: initialization, exploration, and development. Combining basic memory strategies, forgetting and replenishment strategies, and dream-based sharing strategies, it achieves a good balance between global exploration and local development capabilities. It can quickly solve for the optimal torque distribution scheme under the strong disturbance caused by a tire blowout, effectively avoiding local optima problems and significantly improving the robustness and convergence speed of the control system.

[0024] This invention introduces a pure tracking algorithm to achieve lateral steering correction. By pre-aiming at a target point on the reference path, it solves the front wheel steering angle command in real time based on the vehicle's geometric kinematics, actively correcting the vehicle's direction to counteract the tendency to veer after a tire blowout. This control method addresses the problem that vehicles are prone to veering and deviating from the predetermined driving path after a tire blowout, where torque control alone is insufficient to quickly correct the driving direction. By directly solving the steering angle based on kinematic geometry, it can smoothly suppress the sudden trajectory changes caused by a tire blowout and avoid oscillations in control commands. This invention combines a dual-layer torque control architecture with a pure tracking control strategy, simultaneously acting on the vehicle's torque distribution and steering system. It coordinates and regulates both attitude stability and path tracking, achieving a comprehensive improvement in vehicle driving stability after a tire blowout. Attached Figure Description

[0025] Figure 1 A flowchart illustrating the stability control method for vehicles experiencing tire blowouts;

[0026] Figure 2 This is a finite element tire model based on LS-DYNA;

[0027] Figure 3 This is a two-degree-of-freedom model of the vehicle.

[0028] Figure 4 The flowchart is for fuzzy PID control.

[0029] Figure 5 A comparison chart of yaw rates under straight-line tire blowout conditions;

[0030] Figure 6 A comparison chart of the center of gravity sideslip angle under straight-line tire blowout conditions;

[0031] Figure 7 A comparison diagram of lateral displacement under straight-line tire blowout conditions;

[0032] Figure 8 A comparison chart of lateral acceleration under straight-line tire blowout conditions;

[0033] Figure 9 A comparison chart of yaw moment is added for straight-line tire blowout conditions. Detailed Implementation

[0034] The present invention will be further illustrated below with reference to the accompanying drawings and specific embodiments. It should be understood that these embodiments are for illustrative purposes only and are not intended to limit the scope of the invention. After reading this invention, any modifications of the invention in various equivalent forms by those skilled in the art will fall within the scope defined by the appended claims.

[0035] Example 1:

[0036] The present invention provides a vehicle stability control method for a tire blowout, which is executed after a tire blowout is detected, such as... Figure 1 As shown, it includes:

[0037] The vehicle’s current yaw rate and center of gravity sideslip angle are obtained and compared with the corresponding target reference values ​​obtained based on the vehicle dynamics model to obtain the deviation and rate of change of the two state variables.

[0038] The deviation and the rate of change of deviation are input into the fuzzy PID controller, and the additional yaw moment command is generated to suppress vehicle instability by online adaptive adjustment of the PID gain parameter.

[0039] The dream optimization algorithm is used to solve the problem with the four-wheel torque distribution coefficient as the optimization variable. The optimal torque distribution coefficient of the four-wheel independent drive that can realize additional yaw moment is obtained. At the same time, the pure tracking algorithm is used to select the target point at the preset aiming distance in front of the vehicle on the reference path and solve the front wheel steering angle command used to correct the vehicle's driving direction.

[0040] The output torque of the four wheels is controlled by the optimal torque distribution coefficient, and the vehicle steering is controlled by the front wheel steering angle command to achieve vehicle driving stability control after a tire blowout.

[0041] In one embodiment, the target reference value is obtained as follows: a state-space model is established based on the two-degree-of-freedom vehicle dynamics theory, the ideal reference state of the vehicle's lateral dynamics is derived under the steady-state driving assumption, and then the road surface adhesion coefficient is introduced as a physical constraint boundary to correct the limit values ​​of the yaw rate and the center of gravity sideslip angle. The target expected state of the vehicle under the current working condition is calculated, and the target expected state includes the target reference values ​​corresponding to the yaw rate and the center of gravity sideslip angle. Figure 3 This is a two-degree-of-freedom model of the vehicle. In the figure, the horizontal axis represents the vehicle's longitudinal X-axis, and the vertical axis represents the vehicle's lateral Y-axis. The parameters are labeled as follows: a is the distance from the center of gravity to the front axle, b is the distance from the center of gravity to the rear axle, L is the wheelbase, u is the longitudinal speed, v is the lateral speed, and β is the sideslip angle. δ is the yaw rate, and δ is the front wheel steering angle. The lateral force is from the front wheel. The lateral force of the rear wheel is used to solve for the target reference values ​​of yaw rate and center of gravity sideslip angle based on this model.

[0042] In one embodiment, when the fuzzy PID controller is working, the deviation between the actual operating state of the vehicle and the target desired state and the rate of change of the deviation are used as the input of the fuzzy PID. The proportional, integral and derivative gain parameters are adaptively tuned online through preset fuzzy rules, and the output is used to counteract the additional yaw moment for tire blowout disturbance.

[0043] Figure 4 This diagram illustrates the principle of fuzzy PID closed-loop control. The system first inputs the desired yaw rate. This desired reference value is compared with the actual yaw rate collected in real-time from the vehicle dynamics model, and the difference is calculated in the deviation comparison stage to obtain the yaw rate deviation *e*. The differential of the deviation is then used to obtain the yaw rate deviation change rate *ec*. The two sets of state parameters, *e* and *ec*, are used as inputs to the fuzzy controller. These parameters undergo three core processing stages: fuzzification, fuzzy inference, and defuzzification. After calculation, the real-time correction values ​​KP, KI, and KD of the three PID control parameters are output. These correction values ​​are fed into the PID controller to achieve online adaptive tuning of the proportional coefficient KP, integral coefficient KI, and derivative coefficient KD. Based on the real-time updated parameters, the PID controller outputs an additional yaw torque, which is applied to the vehicle dynamics model to output a control quantity that suppresses tire blowout instability.

[0044] Once the KP, KI, and KD parameters of a conventional fixed-parameter PID controller are set, they remain unchanged throughout the entire process. However, when faced with the drastic changes in vehicle dynamic parameters and the strong nonlinearity of the system during a tire blowout, a single fixed parameter is insufficient to simultaneously ensure system response speed, steady-state accuracy, and anti-interference capability. This invention focuses on the goal of vehicle lateral stability control. First, it defines reasonable numerical domains for two input variables—yaw rate deviation and deviation change rate—and three sets of parameters to be corrected: KP, KI, and KD. Then, it uses membership functions adapted to the vehicle control scenario to transform precise input physical quantities into fuzzy linguistic quantities. Next, it builds a complete fuzzy inference rule base based on vehicle instability control experience. According to the vehicle's dynamic response characteristics under different deviation and deviation change conditions, it matches corresponding parameter adjustment logic: when the yaw rate deviation is large, it appropriately increases the proportional coefficient KP to accelerate the system response speed, while limiting the integral coefficient KI to avoid overshoot caused by integral accumulation; when the deviation gradually decreases and the deviation change rate is relatively flat, it slightly decreases KP and reasonably increases KI to eliminate steady-state error; when the deviation changes drastically, it suppresses system oscillations by increasing the derivative coefficient KD, predicts vehicle instability trends, and corrects the control output in advance.

[0045] After completing logical reasoning based on preset fuzzy rules, the controller restores the fuzzy parameter adjustments to precise values ​​that can be directly used in calculations through defuzzification. It iteratively corrects the initial three sets of PID parameters in real time, dynamically updating the KP, KI, and KD values ​​in each control cycle, and continuously outputs the optimal additional yaw moment to constrain the vehicle's driving posture. The vehicle dynamics model provides real-time feedback of the actual yaw rate to the front-end deviation comparison stage, forming a negative feedback closed-loop control. The control output is iteratively corrected by continuously collecting vehicle operating status data from sensors. This invention has undergone repeated iterative debugging under multiple simulation conditions with different vehicle speeds and tire blowout locations, continuously optimizing the membership function range and improving the fuzzy rule constraints, ultimately obtaining a fuzzy PID closed-loop control strategy adapted to vehicle tire blowout instability suppression scenarios.

[0046] In one embodiment, to address the drastic changes and strong nonlinearities in vehicle dynamic parameters at the moment of a tire blowout, and to solve the complex multi-objective optimization problem of real-time torque distribution among the four wheels under blowout conditions, the lower layer of the two-layer collaborative control architecture of this invention employs a novel metaheuristic optimization algorithm inspired by human dreams: the Dream Optimization Algorithm. This algorithm maps vehicle stability control indicators to optimization objectives and addresses the uncertainty caused by a tire blowout by simulating the memory and reorganization mechanisms of human dreams. The process of solving for the optimal torque distribution coefficient using the Dream Optimization Algorithm is divided into three stages: initialization, exploration, and development. It balances global optimization and local search capabilities through basic memory strategies, forgetting and replenishment strategies, and dream-sharing strategies, adapting to the drastic changes in vehicle dynamic parameters after a tire blowout.

[0047] In one embodiment, during the initialization phase, a random population is generated within the search space as an initial solution to initiate the optimization process. Simultaneously, using a basic memory strategy, the optimal control parameters at the moment before the tire blowout are directly assigned to a subset of individuals in the current population as an initial reference solution or as a guide for the initial population. This warm-start approach avoids the algorithm blindly searching from scratch, thereby effectively shortening the control response delay in the initial stage of the blowout.

[0048] Furthermore, the initialization phase will be illustrated in detail below with an example:

[0049] First, a random population is generated within the search space as the initial population to initiate the optimization process of the algorithm. The equation for obtaining the initial population is as follows: Where N represents the population size (number of individuals), and in this embodiment, the preferred value is 8. This value is a balance that, as verified by experiments, maintains basic optimization capability while ensuring that the single-step calculation time of the vehicle controller meets the real-time requirements. This represents the position vector of the i-th individual (i.e., the torque distribution scheme of the four wheels); and These represent the lower and upper boundaries of the search space, respectively, and their values ​​dynamically change with the road adhesion coefficient μ and the wheel condition. For a normal wheel, Set to 300 μ. Set to −150⋅μ; for failed wheels, Set to 20. The setting is 40. This dynamic boundary setting ensures that the torque command output by the algorithm is always within the physical limits of the motor and the mechanical constraints of a tire blowout.

[0050] In one embodiment, during the exploration phase, the algorithm employs a memory-guided differential search strategy for position updates. Specifically, before each iteration, individuals are not randomly reset, but rather guided by excellent solutions from the historical memory pool. A differential perturbation vector is constructed by combining the best and worst solutions of the current population for the search. Through this mechanism, the algorithm can quickly lock the search range using steady-state experience before a tire blowout and dynamically adjust the upper and lower boundaries of torque output based on the real-time road adhesion coefficient (e.g., automatically shrinking the search range on low-adhesion roads). This allows for adaptive searching of the optimal torque combination to suit new tire conditions (such as yaw rate divergence or centroid sideslip angle exceeding limits). Simultaneously, a dynamically decaying learning factor is introduced to balance global exploration and local exploitation capabilities, enhancing the algorithm's ability to escape local optima.

[0051] The following example illustrates the exploration phase in detail:

[0052] First, the entire population is divided into 5 groups based on differences in memory ability (i.e., the number of groups G=5). Each algorithm iteration is regarded as a dream behavior, and the goal is to find the optimal solution. The recommended range for the number of groups G is [3,7], with a preferred value of 5. Too few groups will lead to insufficient population diversity and easy getting trapped in local optima; too many groups will increase the computational burden and reduce the real-time response speed.

[0053] During each dream, an individual randomly selects certain dimensions to update their memory; these dimensions are called forgetting dimensions. The number of forgetting dimensions (parameters) varies among groups with different memory abilities. arrive The forgetting dimension also differs. In this embodiment, the number of forgetting dimensions decreases as the group's memory ability increases, ranging from [1, 4]. For example, the forgetting dimension of the first group, with the weakest memory ability, is [number missing]. =4, the fifth forgetting dimension with the strongest memory capacity. =1. This differentiated setting allows the weak memory group to perform global exploration through large-scale resets, while the strong memory group can retain more historical experience for fine-grained local searches.

[0054] Before each dream sequence, all individuals in the group recall the best individual from the previous iteration. Each individual resets its position information to the position of the best individual in the group, as shown in the following formula:

[0055] ;

[0056] ;

[0057] ;

[0058] ;

[0059] in, The updated individual location; This is the current globally optimal position; Let i be the update component of the i-th individual in the j-th dimension; , … The set of dimension indices to be perturbed; For dream disturbance quantity; from the current position random location The dynamic step size factor α(t) is calculated; rand is a random number in the range [0,1]. α(t) is the dynamic step size factor (innovation parameter), with a value of [0,1]. This parameter decays non-linearly with the operating condition variable t (current iteration number): the initial amplitude is large for global exploration, and it approaches 0 in the later stage for fine-tuning. The maximum number of iterations is 100, with a typical range of [50, 200]. A value below 50 will result in insufficient search, while a value above 200 will increase computation time.

[0060] When an individual is dreaming, they randomly forget some information. Subsequently, the algorithm updates the individual's position based on these forgotten dimensions, combining information from both global and local searches. The update formula is as follows.

[0061] To further enhance the ability to escape local optima, the algorithm introduces an information-sharing mechanism. When an individual updates its own forgetting dimension, it has a probability of randomly obtaining information about other individuals in the same group on the corresponding dimension. ;in, This is the update component for the m-th individual; is the current iteration position of the m-th individual; N is the total population size.

[0062] In one embodiment, during the development phase, all individuals in the population share the global optimal solution. The position information of the group's optimal solution is retained in the non-forgetting dimension to quickly move towards the optimal solution. In the forgetting dimension, the position is updated locally through random perturbation and cosine function self-organization to achieve fine-grained search. At the same time, the torque distribution among the four wheels is coordinated through a dream-sharing strategy to avoid single-wheel torque exceeding the limit or locking, ensuring rapid convergence of yaw rate and center of mass sideslip angle under extreme working conditions.

[0063] The following example illustrates the development phase in detail:

[0064] Once the algorithm enters the development phase, grouping operations cease. At this point, all individuals in the population share the best solution from the previous iteration (i.e., the global optimum). Before each dream begins, each individual updates its position information on the forgetting dimension. This phase also employs three strategies: basic memory, forgetting and replenishment, and dream sharing.

[0065] The basic memory strategy is shown in the following formula. In the non-forgetting dimension, individuals can remember and retain the position information of the group's optimal solution, thereby quickly moving towards the global optimal solution. ;

[0066] Forgetting and replenishment strategy: As shown in the following formula, on the forgetting dimension, individuals update their positions through random perturbation and cosine function self-organization to achieve local fine search and avoid getting trapped in local optima.

[0067] ;

[0068] ;

[0069] ;

[0070] Among them, is the updated position of the i-th individual in the j-th dimension; is the updated component of m individuals; is the current iteration position of the m-th individual; N is the total population size; , … is a randomly selected set of dimension indices used to determine the specific dimensions for position update.

[0071] In one embodiment, (to verify the effectiveness of the method of the present invention, it is necessary to calibrate and initialize the key parameters of the algorithm through a large number of numerical experiments. Combining the analysis of algorithm stability and applicability, the core parameters are determined as follows: the number of iterations in the exploration stage As shown, the maximum number of iterations in the exploration stage is set to nine-tenths of the total maximum number of iterations . This ensures that the algorithm explores for most of the time to discover potential high-quality solutions, ; among them, represents the maximum number of iterations in the exploration stage, represents the total maximum number of iterations.

[0072] A relatively large exploration ratio (i.e., the coefficient is close to 0.9) is set to ensure that the algorithm is in the global exploration state for most of the time, fully searching the solution space to discover potential high-quality solution regions; if this coefficient is too low (e.g., less than 0.7), it will cause the algorithm to enter the exploitation stage prematurely and be prone to falling into local optima; if it is too high (e.g., greater than 0.95), it will compress the time for later fine search, resulting in insufficient convergence accuracy.

[0073] The number of forgotten dimensions and are as shown in the following formula. The number of forgotten dimensions is determined by the problem dimension Dim. For the q-th group of individuals in the exploration stage, the number of its forgotten dimensions is an integer randomly selected from a specific range; for the exploitation stage, the number of forgotten dimensions is also determined by a similar rule. This dynamic setting enables the algorithm to adapt to problems of different dimensions, ; ;

[0074] Set the parameter u = 0.9. When Dim < u, execute the forgetting supplement strategy; otherwise, execute the dream sharing strategy.

[0075] To avoid the individual position exceeding the search space boundary during the iteration process, for optimization problems of different dimensions, two different boundary handling strategies are designed. For low-dimensional problems with fewer local optima, the traditional random reset method is used to regenerate the out-of-bounds dimension randomly within the valid range, as shown in the following formula: ;in, , are the upper and lower boundaries of the j-th dimension, respectively, and rand is a random number between 0 and 1.

[0076] High-dimensional problems are complex and have more local optima. To enhance global optimization capabilities, a strategy similar to dream sharing is adopted, reusing the effective positions of other individuals in the population to update out-of-bounds dimensions, as shown in the following equation: Where m is an individual index randomly selected from the population that is different from i, ensuring that the updated position is still within the valid search space.

[0077] Both strategies update only the dimensions that exceed the boundary, ensuring that all individuals remain within the feasible region while avoiding excessive perturbation of effective solutions. For the torque optimization allocation requirement in the tire blowout scenario of distributed-drive electric vehicles, the population size of the dream optimization algorithm is set to 8, the total number of iterations to 150, and a 4-dimensional particle dimension is selected to correspond to the independent torque allocation of the four wheels. The maximum number of iterations in the exploration phase is set to nine-tenths of the total maximum number of iterations. The number of forgotten dimensions is dynamically set according to the problem dimension, and differentiated boundary handling strategies are adopted for high- and low-dimensional problems to ensure that the individual position always remains within the feasible search space.

[0078] In one embodiment, when using a pure tracking algorithm to solve for the front wheel steering angle command, a point on the reference path at a preset aiming distance from the center point of the vehicle's rear axle is first selected as the tracking target. Based on the geometric relationship between the vehicle's current position, heading, and the target point, the correspondence between the turning radius, aiming distance, and yaw angle is determined using the sine theorem, and the trajectory curvature is calculated. Then, based on the Ackermann steering geometry principle and the vehicle's wheelbase, the relationship between the front wheel steering angle and curvature is established, and finally, the desired front wheel steering angle that allows the vehicle to smoothly follow the predetermined path is solved.

[0079] Furthermore, the following example illustrates the specific process of the pure tracking algorithm solving for the front wheel steering angle command:

[0080] First, based on the principles of vehicle geometry and kinematics, during vehicle movement, a reference path is selected with a distance of [distance from the rear axle center point] as [value]. The preset forward point is taken as the current tracking target; then, based on the geometric relationship between the vehicle's current position, heading and the target point, the functional relationship between the direction angle of the line connecting the rear axle of the vehicle to the path point Q, the vehicle's yaw angle ψ and the front wheel steering angle η is established.

[0081] Specifically, the derivation process of the functional relationship is as follows:

[0082] Establish geometric relationships: Use the sine theorem to determine the turning radius R and the pre-aiming distance. The relationship between the yaw angle ψ and the yaw angle ψ is: ;in, : Preview distance, representing the longitudinal distance from the center point of the vehicle's rear axle to the target point ahead of the path; ψ represents the vehicle's yaw angle, which is the angle between the vehicle's heading and the tangent direction of the reference path; R represents the vehicle's current turning radius.

[0083] Calculating the trajectory curvature: Based on the above geometric relationships, the formula for calculating the curvature γ of the turning trajectory is as follows: ;

[0084] Where γ represents the curvature of the vehicle's current trajectory;

[0085] Determining the front wheel steering angle: Based on the Ackermann steering geometry principle, combined with the wheelbase... The relationship between the front wheel steering angle and curvature is established as follows: ;in, η represents the vehicle wheelbase, which is the distance between the center points of the front and rear axles; η represents the front wheel steering angle, which is the angle of deflection of the front wheel relative to the longitudinal axis of the vehicle body.

[0086] Finally, by combining the above formulas, we obtain the control law for the desired front wheel steering angle: ;in, This represents the desired front wheel steering angle command for the vehicle at time t.

[0087] By following the steps above, the front wheel steering angle required for the vehicle's rear axle center to travel along an arc trajectory passing through the pre-aiming point is calculated, thereby achieving smooth following of the predetermined path.

[0088] To verify the control effect of the combined control strategy of this invention under the straight-line tire blowout condition, four schemes were set up for simulation comparison: no control, FPID+DOA control, pure tracking control, and the combined control of this invention. The simulation results are as follows: Figure 5 , Figure 6 , Figure 7 , Figure 8 , Figure 9 As shown. Legendary meanings: Black curve (no control), red curve (FPID+DOA), blue curve (pure tracking), green curve (joint control), purple curve (reference value). The horizontal axis of all sub-graphs represents time (unit: seconds).

[0089] Figure 5 The vertical axis represents the yaw rate (unit: deg / s). Under the combined control of this invention, the peak yaw rate is the lowest and can quickly converge to the reference value, effectively suppressing vehicle yaw instability. Figure 6 The vertical axis represents the centroid sideslip angle (unit: deg). The joint control strategy can quickly correct the centroid sideslip angle to the ideal reference range and suppress vehicle body sideslip. Figure 7The vertical coordinate represents the lateral displacement (unit: m). Compared with other control schemes, the present invention has the smallest lateral offset and the best path-keeping capability. Figure 8 The vertical axis represents lateral acceleration (unit: g). The joint control strategy effectively reduces lateral acceleration overshoot and improves vehicle ride comfort. Figure 9 The vertical axis represents the additional yaw moment (unit: N). This invention can quickly output the optimal additional yaw moment to counteract tire blowout disturbances, and the moment convergence speed is better than other control strategies.

[0090] In one embodiment, the method of the present invention further includes an offline step of pre-constructing a blowout tire model: using LS-DYNA explicit dynamics software to establish a tire finite element model, simulating the tire inflation and load-bearing process by applying air pressure load to the inner surface of the tire and contact constraints with the moving road surface, and simulating and analyzing the nonlinear decay law of the tire's longitudinal stiffness and lateral stiffness under the blowout condition, then constructing a vehicle dynamics model based on the UniTire tire model theory, and correcting the key parameters of the UniTire model according to the stiffness decay law, and integrating the blowout tire model characterizing the mechanical properties after the blowout into the vehicle dynamics model. Figure 2 This is a finite element tire model based on LS-DYNA. Based on this finite element model, tire inflation loading and road contact simulation can be completed, and tire mechanical parameters under the condition of tire blowout can be obtained, providing data support for the parameter correction of UniTire whole vehicle tire model.

[0091] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for controlling vehicle stability after a tire blowout, executed upon detection of a tire blowout, characterized in that, include: The vehicle’s current yaw rate and center of gravity sideslip angle are obtained and compared with the corresponding target reference values ​​obtained based on the vehicle dynamics model to obtain the deviation and rate of change of the two state variables. The deviation and the rate of change of deviation are input into the fuzzy PID controller, and the additional yaw moment command for suppressing vehicle instability is generated by online adaptive adjustment of the PID gain parameter. The dream optimization algorithm is used to solve the problem with the four-wheel torque distribution coefficient as the optimization variable to obtain the optimal torque distribution coefficient for four-wheel independent drive that can realize the additional yaw moment. At the same time, the pure tracking algorithm is used to select the target point at a preset aiming distance in front of the vehicle on the reference path and solve the front wheel steering angle command used to correct the vehicle's driving direction. The output torque of the four wheels is controlled according to the optimal torque distribution coefficient, and the vehicle steering is controlled according to the front wheel steering angle command to achieve vehicle driving stability control after a tire blowout.

2. The method for controlling vehicle stability after a tire blowout according to claim 1, characterized in that, The target reference value is obtained as follows: a state-space model is established based on the two-degree-of-freedom vehicle dynamics theory, the ideal reference state of the vehicle's lateral dynamics is derived under the steady-state driving assumption, and then the road surface adhesion coefficient is introduced as a physical constraint boundary to correct the limit values ​​of yaw rate and center of gravity sideslip angle. The target expected state of the vehicle under the current working condition is calculated, and the target expected state includes the target reference values ​​corresponding to the yaw rate and center of gravity sideslip angle.

3. The method for controlling vehicle stability after a tire blowout according to claim 1, characterized in that, When the fuzzy PID controller is working, the deviation between the actual operating state of the vehicle and the target desired state and the rate of change of the deviation are used as the input of the fuzzy PID controller. The proportional, integral and derivative gain parameters are adaptively tuned online through preset fuzzy rules, and the output is used to counteract the additional yaw moment of the tire blowout disturbance.

4. The method for controlling vehicle stability after a tire blowout according to claim 1, characterized in that, The process of solving the optimal torque distribution coefficient using the dream optimization algorithm is divided into three stages: initialization, exploration, and development. The algorithm balances global optimization and local search capabilities through basic memory strategy, forgetting and supplementation strategy, and dream sharing strategy.

5. The method for controlling vehicle stability after a tire blowout according to claim 4, characterized in that, During the initialization phase, a random population is generated in the search space as an initial solution to start the optimization process; at the same time, the basic memory strategy is used to call the steady-state control experience before the tire blowout as part of the initial solution of the initial population to shorten the control response delay in the early stage of the tire blowout.

6. The method for controlling vehicle stability after a tire blowout according to claim 4, characterized in that, During the exploration phase, the population was divided into multiple groups based on differences in memory capacity. Before each iteration considered as dream behavior, individuals within the group were reset to their historical best positions. Then, some dimensions were randomly selected as forgetting dimensions for position updates. Through forgetting and replenishment strategies, old parameter constraints that failed due to tire blowouts were dynamically eliminated. The algorithm adaptively searched for the optimal torque combination that matched the new tire state. At the same time, an information sharing mechanism within the group was introduced to enhance the algorithm's ability to escape local optima.

7. The method for controlling vehicle stability after a tire blowout according to claim 4, characterized in that, During the development phase, all individuals in the population share the global optimal solution. In the non-forgetting dimension, the position information of the group's optimal solution is retained to quickly move towards the optimal solution. In the forgetting dimension, the position is updated through random perturbation and cosine function self-organization to achieve local fine search. At the same time, the torque distribution among the four wheels is coordinated through the dream sharing strategy.

8. The method for controlling vehicle stability after a tire blowout according to claim 4, characterized in that, To address the torque optimization and allocation requirements of distributed drive electric vehicles under tire blowout conditions, the population size of the dream optimization algorithm is set to 8, the total number of iterations is set to 150, and a 4-dimensional particle dimension is selected to correspond to the independent torque allocation of the four wheels. The maximum number of iterations in the exploration phase is set to nine-tenths of the total maximum number of iterations. The number of forgotten dimensions is dynamically set according to the problem dimension, and differentiated boundary handling strategies are adopted for high- and low-dimensional problems.

9. The method for controlling vehicle stability after a tire blowout according to claim 1, characterized in that, When using a pure tracking algorithm to solve for the front wheel steering angle command, the following steps are taken: First, a point on the reference path at a distance of a preset aiming distance from the center point of the vehicle's rear axle is selected as the tracking target. Based on the geometric relationship between the vehicle's current position, heading, and the target point, the correspondence between the turning radius, aiming distance, and yaw angle is determined using the sine theorem, and the trajectory curvature is calculated. Then, based on the Ackermann steering geometry principle and the vehicle's wheelbase, the relationship between the front wheel steering angle and curvature is established. Finally, the desired front wheel steering angle that allows the vehicle to smoothly follow the predetermined path is solved.

10. The method for controlling vehicle stability after a tire blowout according to claim 1, characterized in that, The method also includes an offline step of pre-constructing a blowout tire model: using LS-DYNA explicit dynamics software to establish a tire finite element model, simulating the tire inflation and load-bearing process by applying air pressure load to the inner surface of the tire and contact constraints with the moving road surface, and simulating and analyzing the nonlinear decay law of the tire's longitudinal stiffness and lateral stiffness under the blowout condition, then constructing a vehicle dynamics model based on the UniTire tire model theory, and correcting the key parameters of the UniTire model according to the stiffness decay law, and integrating the blowout tire model characterizing the mechanical properties after the blowout into the vehicle dynamics model.

Images