Parameter design method and device of active ISD suspension

Abstract

The application relates to a parameter design method and device of an active ISD suspension, and belongs to the technical field of vehicle suspension control, wherein the parameter design method of the active ISD suspension comprises the following steps: constructing a vehicle dynamics model of an active ISD suspension system and an MPC controller framework of the active ISD suspension system based on a sprung mass, an unsprung mass, a tire equivalent spring stiffness, a driving speed and a road roughness coefficient of a target vehicle; determining a damping coefficient, a spring stiffness coefficient and a mass inertia coefficient of each three-element unit based on a GA algorithm and the vehicle dynamics model; and determining input weights and output weights of the active ISD suspension system based on the determined damping coefficient, the spring stiffness coefficient and the mass inertia coefficient of each three-element unit, an MPC controller framework and an RL model. The application can improve the parameter design efficiency of the active ISD suspension while ensuring the vehicle performance.

Description

Technical Field

[0001] This invention relates to the field of vehicle suspension control technology, and in particular to a parameter design method and device for an active ISD suspension. Background Technology

[0002] The vehicle suspension system is a core component affecting ride comfort, handling stability, and passenger comfort. Traditional passive suspensions have fixed parameters, making it difficult to adapt to complex and changing road conditions.

[0003] In recent years, inertia-spring-damper (ISD) suspensions have attracted widespread attention as a novel passive network structure. An inertia is a passive mechanical element with two endpoints, whose output force is proportional to the relative acceleration between those endpoints. Theoretical analysis and experiments show that the appropriate introduction of an inertia into the suspension can effectively improve vibration isolation performance at low frequencies. To further enhance performance, combining ISD suspensions with active control to form active ISD suspensions has become a current research frontier. However, existing active ISD suspension engineering design schemes struggle to guarantee vehicle performance, and the high computational complexity of the design process leads to low design efficiency.

[0004] Therefore, how to improve the efficiency of active ISD suspension parameter design and improve vehicle performance through active ISD suspension parameter design has become an urgent technical problem to be solved. Summary of the Invention

[0005] In view of this, it is necessary to provide a parameter design method and device for active ISD suspension to solve the problems of low efficiency and difficulty in guaranteeing vehicle performance of existing active ISD suspension parameter design schemes.

[0006] To address the aforementioned problems, in a first aspect, the present invention provides a parameter design method for an active ISD suspension, comprising:

[0007] Based on the sprung mass, unsprung mass, tire equivalent spring stiffness, vehicle speed, and road roughness coefficient of the target vehicle, a vehicle dynamics model and an MPC controller framework for the active ISD suspension system are constructed. The active ISD suspension system is composed of a first three-element unit connected in series with a second three-element unit and then connected in parallel with a third three-element unit. Each three-element unit contains a damping element, a spring element, and an inertial capacitance element. The vehicle dynamics model is used to solve for the damping coefficient, spring stiffness coefficient, and inertial mass coefficient of each three-element unit. The MPC controller framework is used to solve for the input weights and output weights of the active ISD suspension system.

[0008] The damping coefficient, spring stiffness coefficient, and inertia coefficient of each three-element unit are determined based on the GA algorithm and vehicle dynamics model. Based on the determined damping coefficient, spring stiffness coefficient, inertia coefficient, MPC controller frame, and RL model of each three-element unit, the input weights and output weights of the active ISD suspension system are determined. The individual fitness of the GA algorithm is determined based on the sprung mass acceleration, suspension dynamic travel, and tire dynamic deflection of the active ISD suspension system. During training, the RL model uses multiple sets of sample damping coefficients, sample spring stiffness coefficients, and sample inertia coefficients determined by the GA algorithm as samples, the input weights and output weights of the active ISD suspension system as the action space, and the reciprocal of the individual fitness of the GA algorithm as the reward function.

[0009] In one possible implementation, the vehicle dynamics model is represented by the following formula:

[0010]

[0011]

[0012]

[0013] in, Indicates unsprung mass. Indicates the sprung mass. This indicates the equivalent spring stiffness of the tire. Indicates power. Indicates single-wheel road surface excitation. Represents the vertical displacement of the unsprung mass. Represents the vertical velocity of the unsprung mass. Represents the vertical acceleration of the unsprung mass. This represents the vertical displacement of the intermediate connection point. This indicates the vertical velocity at the intermediate connection point. This represents the vertical acceleration at the intermediate connection point. Represents the vertical displacement of the sprung mass. Represents the vertical velocity of the sprung mass. Represents the vertical acceleration of the sprung mass. This represents the damping coefficient of the first three-element unit. This represents the spring stiffness coefficient of the first three-element unit. This represents the mass inertia coefficient of the first three-element unit. This represents the damping coefficient of the second and third element units. This represents the spring stiffness coefficient of the second and third element units. This represents the mass inertia coefficient of the second and third element units. This represents the damping coefficient of the third element unit. This represents the spring stiffness coefficient of the third element unit. This represents the mass inertia coefficient of the third element unit.

[0014] In one possible implementation, the single-wheel road surface excitation is determined based on the following formula:

[0015]

[0016] in, express The first derivative with respect to time, Indicates single-wheel road surface excitation. Indicates the reference space frequency. Indicates vehicle speed. This represents the road surface roughness coefficient. This represents ideal white noise.

[0017] In one possible implementation, the MPC controller framework is represented by the following formula:

[0018]

[0019]

[0020]

[0021]

[0022]

[0023] in, This represents the state vector of the active ISD suspension system. Indicates the input of the active ISD suspension system. This indicates the output of the active ISD suspension system. express The first derivative with respect to time, Indicates power. Indicates single-wheel road surface excitation. Represents the vertical displacement of the unsprung mass. Represents the vertical velocity of the unsprung mass. This represents the vertical displacement of the intermediate connection point. This indicates the vertical velocity at the intermediate connection point. Represents the vertical displacement of the sprung mass. Represents the vertical velocity of the sprung mass. , , , This is the system matrix for the MPC controller framework.

[0024] In one possible implementation, the individual fitness of the GA algorithm is determined based on the following formula:

[0025]

[0026] in, Indicates individual fitness. Indicates the acceleration of the sprung mass. Indicates the suspension travel. Indicates tire deflection. , , These are the preset normalized weighting coefficients.

[0027] In one possible implementation, the convergence condition of the GA algorithm includes:

[0028] The fitness values ​​of the GA population for 10 consecutive generations fluctuate within a range less than the fluctuation threshold, and the root mean square value of the sprung mass acceleration is less than or equal to the first threshold, the root mean square value of the suspension dynamic travel is less than or equal to the second threshold, and the root mean square value of the tire dynamic deflection is less than or equal to the third threshold.

[0029] In one possible implementation, the method further includes:

[0030] Perform a structural rationality check on each individual in the GA population of each generation of the GA algorithm, and discard individuals that fail the structural rationality check;

[0031] If a target individual meets at least one of the following conditions, it is determined that the target individual failed the structural rationality test. The target individual can be any individual in any generation of the GA population:

[0032] The damping coefficient, spring stiffness coefficient, and mass inertia coefficient of the first three-element unit are all 0, or the damping coefficient, spring stiffness coefficient, and mass inertia coefficient of the second three-element unit are all 0.

[0033] The damping coefficient and inertia coefficient of the first three-element unit are both 0, the damping coefficient and inertia coefficient of the second three-element unit are both 0, and the spring stiffness coefficient of the first three-element unit and the spring stiffness coefficient of the second three-element unit are both not 0.

[0034] The mass inertia coefficients of both the first and second three-element units are 0.

[0035] The spring stiffness coefficient of the third three-element unit is 0, and at least one of the spring stiffness coefficients of the first three-element unit and the second three-element unit is 0.

[0036] On the other hand, the present invention also provides a parameter design device for an active ISD suspension, comprising:

[0037] The module is used to construct the vehicle dynamics model and MPC controller framework of the active ISD suspension system based on the sprung mass, unsprung mass, tire equivalent spring stiffness, vehicle speed, and road roughness coefficient of the target vehicle. The active ISD suspension system is composed of a first three-element unit connected in series with a second three-element unit and then connected in parallel with a third three-element unit. Each three-element unit contains a damping element, a spring element, and an inertial capacitance element. The vehicle dynamics model is used to solve for the damping coefficient, spring stiffness coefficient, and inertial mass coefficient of each three-element unit. The MPC controller framework is used to solve for the input weights and output weights of the active ISD suspension system.

[0038] The determination module is used to determine the damping coefficient, spring stiffness coefficient, and inertia coefficient of each three-element unit based on the GA algorithm and vehicle dynamics model. Based on the determined damping coefficient, spring stiffness coefficient, inertia coefficient, MPC controller frame, and RL model of each three-element unit, the input weights and output weights of the active ISD suspension system are determined. The individual fitness of the GA algorithm is determined based on the sprung mass acceleration, suspension dynamic travel, and tire dynamic deflection of the active ISD suspension system. During training, the RL model uses multiple sets of sample damping coefficients, sample spring stiffness coefficients, and sample inertia coefficients determined by the GA algorithm as samples, the input weights and output weights of the active ISD suspension system as the action space, and the reciprocal of the individual fitness of the GA algorithm as the reward function.

[0039] Secondly, the present invention also provides a parameter design device, including a memory and a processor, wherein,

[0040] The memory is used to store programs;

[0041] The processor, coupled to the memory, is used to execute the program stored in the memory to implement the steps in the parameter design method for the active ISD suspension described in any of the above implementations.

[0042] Thirdly, the present invention also provides a computer-readable storage medium for storing a computer-readable program or instructions, which, when executed by a processor, can implement the steps in the parameter design method for the active ISD suspension described in any of the above implementations.

[0043] The beneficial effects of this invention are as follows: The parameter design method and apparatus for active ISD suspension provided by this invention provide theoretical support for the parameter design of active ISD suspension by constructing a vehicle dynamics model and an MPC controller framework. By combining the MPC controller framework with the vehicle dynamics model, the parameter design is transformed into a quadratic programming process to determine the optimal parameters, thereby ensuring that the designed parameters can improve vehicle performance. Then, the active ISD suspension parameters are designed through the GA algorithm and the RL model. The collaboration between the GA algorithm and the RL model improves the iteration efficiency of the GA algorithm, thereby improving the efficiency of parameter design. This invention improves the efficiency of active ISD suspension parameter design while ensuring vehicle performance. Attached Figure Description

[0044] Figure 1 A schematic flowchart of an embodiment of the parameter design method for the active ISD suspension provided by the present invention;

[0045] Figure 2 A schematic diagram of an embodiment of the active ISD suspension mechanical structure provided by the present invention;

[0046] Figure 3 A schematic diagram of an embodiment of the active ISD suspension quarter-vehicle dynamics model provided by the present invention;

[0047] Figure 4 A schematic flowchart illustrating an embodiment of the overall parameter design process for the active ISD suspension provided by the present invention;

[0048] Figure 5 A schematic flowchart of an embodiment of the training phase for parameter design of the active ISD suspension provided by the present invention;

[0049] Figure 6 A schematic flowchart illustrating an embodiment of the application phase of the parameter design for the active ISD suspension provided by the present invention;

[0050] Figure 7 A schematic diagram of an embodiment of the parameter design device for the active ISD suspension provided by the present invention;

[0051] Figure 8 A schematic diagram of an embodiment of the parameter design device provided by the present invention. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0053] In the description of the embodiments of the present invention, unless otherwise stated, "multiple" means two or more. "And / or" describes the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0054] The terms "first," "second," etc., used in the embodiments of this invention are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, a technical feature defined with "first" or "second" may explicitly or implicitly include at least one of that feature.

[0055] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of the invention. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0056] This invention provides a parameter design method and apparatus for an active ISD suspension, which will be described below.

[0057] Figure 1 A schematic flowchart of an embodiment of the parameter design method for the active ISD suspension provided by the present invention is shown below. Figure 1 As shown, the parameter design method for active ISD suspension includes:

[0058] S101. Based on the sprung mass, unsprung mass, tire equivalent spring stiffness, vehicle speed, and road roughness coefficient of the target vehicle, construct the vehicle dynamics model of the active ISD suspension system and the MPC controller framework of the active ISD suspension system. The active ISD suspension system is composed of a first three-element unit connected in series with a second three-element unit and then connected in parallel with a third three-element unit. Each three-element unit contains a damping element, a spring element, and an inertial capacitance element. The vehicle dynamics model is used to solve for the damping coefficient, spring stiffness coefficient, and inertial mass coefficient of each three-element unit. The MPC controller framework is used to solve for the input weights and output weights of the active ISD suspension system.

[0059] It should be noted that the parameter design method for active ISD suspension provided by this invention can be applied to the parameter design scenario of vehicle ISD suspension, especially the parameter design scenario of vehicle active ISD suspension.

[0060] Before designing the parameters of the active ISD suspension, a vehicle dynamics model and a model predictive control (MPC) framework for the active ISD suspension system can be constructed based on the target vehicle's sprung mass, unsprung mass, tire equivalent spring stiffness, vehicle speed, and road roughness coefficient. The vehicle dynamics model can be used to solve for the damping coefficient, spring stiffness coefficient, and inertia coefficient of each three-element unit, while the MPC controller framework can be used to solve for the input and output weights of the active ISD suspension system. By constructing the vehicle dynamics model and the MPC controller framework, theoretical support can be provided for the parameter design of the active ISD suspension, thereby ensuring that the designed parameters can improve vehicle performance.

[0061] Combination Figure 2 The active ISD suspension system consists of a first three-element unit (composed of...) , , Composition) and the second three-element unit (composed of) , , After being connected in series with the third element (composed of...), it is then connected to the third element (composed of...). , , The components are arranged in parallel, and each three-element unit contains a damping element. ), a spring element ( ) and an inertial capacitive element ( ).

[0062] S102. Based on the GA algorithm and vehicle dynamics model, determine the damping coefficient, spring stiffness coefficient, and inertia coefficient of each three-element unit. Based on the determined damping coefficient, spring stiffness coefficient, inertia coefficient, MPC controller frame, and RL model of each three-element unit, determine the input weights and output weights of the active ISD suspension system. The individual fitness of the GA algorithm is determined based on the sprung mass acceleration, suspension dynamic travel, and tire dynamic deflection of the active ISD suspension system. During training, the RL model uses multiple sets of sample damping coefficients, sample spring stiffness coefficients, and sample inertia coefficients determined by the GA algorithm as samples, the input weights and output weights of the active ISD suspension system as the action space, and the reciprocal of the individual fitness of the GA algorithm as the reward function.

[0063] It should be noted that after constructing the vehicle dynamics model and the MPC controller framework, the damping coefficient, spring stiffness coefficient, and inertia coefficient of each three-element unit can first be determined using a genetic algorithm (GA) and the vehicle dynamics model. The individual fitness of the GA algorithm during the iteration process can be determined based on the sprung mass acceleration, suspension travel, and tire deflection of the active ISD suspension system. Then, based on the determined damping coefficient, spring stiffness coefficient, inertia coefficient of each three-element unit, the MPC controller framework, and the reinforcement learning (RL) model, the input and output weights of the active ISD suspension system are determined. During training, the RL model uses multiple sets of sample damping coefficients, sample spring stiffness coefficients, and sample inertia coefficients determined by the GA algorithm as samples, the input and output weights of the active ISD suspension system as the action space, and the reciprocal of the individual fitness of the GA algorithm as the reward function. Using the GA algorithm and the RL model for parameter design of the active ISD suspension can effectively improve the efficiency of parameter design.

[0064] In summary, the parameter design method for active ISD suspension provided by this invention provides theoretical support for the parameter design of active ISD suspension by constructing a vehicle dynamics model and an MPC controller framework. By combining the MPC controller framework with the vehicle dynamics model, the parameter design is transformed into a quadratic programming process to determine the optimal parameters, thereby ensuring that the designed parameters can improve vehicle performance. Then, the active ISD suspension parameters are designed using the GA algorithm and the RL model. The collaboration between the GA algorithm and the RL model improves the iteration efficiency of the GA algorithm, thereby improving the efficiency of parameter design. This invention improves the efficiency of active ISD suspension parameter design while ensuring vehicle performance.

[0065] In some embodiments of the present invention, the vehicle dynamics model is expressed based on the following formula:

[0066]

[0067]

[0068]

[0069] in, Indicates unsprung mass. Indicates the sprung mass. This indicates the equivalent spring stiffness of the tire. Indicates power. Indicates single-wheel road surface excitation. Represents the vertical displacement of the unsprung mass. Represents the vertical velocity of the unsprung mass. Represents the vertical acceleration of the unsprung mass. This represents the vertical displacement of the intermediate connection point. This indicates the vertical velocity at the intermediate connection point. This represents the vertical acceleration at the intermediate connection point. Represents the vertical displacement of the sprung mass. Represents the vertical velocity of the sprung mass. Represents the vertical acceleration of the sprung mass. This represents the damping coefficient of the first three-element unit. This represents the spring stiffness coefficient of the first three-element unit. This represents the mass inertia coefficient of the first three-element unit. This represents the damping coefficient of the second and third element units. This represents the spring stiffness coefficient of the second and third element units. This represents the mass inertia coefficient of the second and third element units. This represents the damping coefficient of the third element unit. This represents the spring stiffness coefficient of the third element unit. This represents the mass inertia coefficient of the third element unit.

[0070] It should be noted that: combination Figure 3 As can be seen, the constructed vehicle dynamics model is a 1 / 4 vehicle dynamics model, and its dynamic equations can be expressed by the above formulas.

[0071] In some embodiments of the present invention, the single-wheel road surface excitation is determined based on the following formula:

[0072]

[0073] in, express The first derivative with respect to time, Indicates single-wheel road surface excitation. Indicates the reference space frequency. Indicates vehicle speed. This represents the road surface roughness coefficient. This represents ideal white noise.

[0074] It should be noted that the single-wheel road surface excitation in the vehicle dynamics model can be determined using the above formula.

[0075] In some embodiments of the present invention, in one possible implementation, the MPC controller framework is represented based on the following formula:

[0076]

[0077]

[0078]

[0079]

[0080]

[0081] in, This represents the state vector of the active ISD suspension system. Indicates the input of the active ISD suspension system. This indicates the output of the active ISD suspension system. express The first derivative with respect to time, Indicates power. Indicates single-wheel road surface excitation. Represents the vertical displacement of the unsprung mass. Represents the vertical velocity of the unsprung mass. This represents the vertical displacement of the intermediate connection point. This indicates the vertical velocity at the intermediate connection point. Represents the vertical displacement of the sprung mass. Represents the vertical velocity of the sprung mass. , , , This is the system matrix for the MPC controller framework.

[0082] It should be noted that the above formula can be used to construct the MPC controller framework.

[0083] In some embodiments of the present invention, the individual fitness of the GA algorithm is determined based on the following formula:

[0084]

[0085] in, Indicates individual fitness. Indicates the acceleration of the sprung mass. Indicates the suspension travel. Indicates tire deflection. , , These are the preset normalized weighting coefficients.

[0086] It should be noted that the individual fitness of the GA algorithm can be determined using the formula above.

[0087] In some embodiments of the present invention, the convergence conditions of the GA algorithm include:

[0088] The fitness values ​​of the GA population for 10 consecutive generations fluctuate within a range less than the fluctuation threshold, and the root mean square value of the sprung mass acceleration is less than or equal to the first threshold, the root mean square value of the suspension dynamic travel is less than or equal to the second threshold, and the root mean square value of the tire dynamic deflection is less than or equal to the third threshold.

[0089] It should be noted that the convergence condition of the GA algorithm can be: the fitness value of the GA population fluctuates within a range less than a fluctuation threshold (e.g., 0.001) for 10 consecutive generations, and the root mean square value of the sprung mass acceleration is less than or equal to a first threshold (e.g., ...). The root mean square value of the suspension dynamic travel is less than or equal to the second threshold (e.g., The root mean square value of tire dynamic deflection is less than or equal to the third threshold (e.g., ).

[0090] In some embodiments of the present invention, the method further includes:

[0091] Perform a structural rationality check on each individual in the GA population of each generation of the GA algorithm, and discard individuals that fail the structural rationality check;

[0092] If a target individual meets at least one of the following conditions, it is determined that the target individual failed the structural rationality test. The target individual can be any individual in any generation of the GA population:

[0093] The damping coefficient, spring stiffness coefficient, and mass inertia coefficient of the first three-element unit are all 0, or the damping coefficient, spring stiffness coefficient, and mass inertia coefficient of the second three-element unit are all 0.

[0094] The damping coefficient and inertia coefficient of the first three-element unit are both 0, the damping coefficient and inertia coefficient of the second three-element unit are both 0, and the spring stiffness coefficient of the first three-element unit and the spring stiffness coefficient of the second three-element unit are both not 0.

[0095] The mass inertia coefficients of both the first and second three-element units are 0.

[0096] The spring stiffness coefficient of the third three-element unit is 0, and at least one of the spring stiffness coefficients of the first three-element unit and the second three-element unit is 0.

[0097] It should be noted that, to ensure the rationality of the active ISD suspension parameter design, a structural rationality check must be performed on each individual in each generation of the GA algorithm population, and individuals that fail the structural rationality check must be discarded. An individual can be considered to have failed the structural rationality check if it meets at least one of the following conditions:

[0098] 1. The damping coefficient, spring stiffness coefficient, and mass inertia coefficient of the first three-element unit are all 0, or the damping coefficient, spring stiffness coefficient, and mass inertia coefficient of the second three-element unit are all 0. In this case, the active ISD suspension system does not meet the dual-layer condition.

[0099] 2. The damping coefficient and inertia coefficient of the first three-element unit are both 0, and the damping coefficient and inertia coefficient of the second three-element unit are both 0. The spring stiffness coefficients of both the first and second three-element units are not 0. In this case, the series connection of the active ISD suspension system only contains springs, which are of no practical significance.

[0100] 3. The inertia coefficients of both the first and second three-element units are 0. In this case, the MPC controller calculation contains terms with a denominator of 0, which is unreasonable.

[0101] 4. The spring stiffness coefficient of the third three-element unit is 0, and at least one of the spring stiffness coefficients of the first and second three-element units is 0. In this case, there is no continuous spring between the sprung mass and the unsprung mass, the support force is interrupted, and the suspension cannot be guaranteed to have effective support capacity at all times during operation.

[0102] This invention establishes a GA–RL two-layer optimization framework. The upper layer optimizes suspension structural parameters, and the lower layer solves for the optimal control strategy under a given structure. This framework includes a training phase and an application phase, such as... Figure 4 As shown. The training phase is as follows. Figure 5 As shown, dedicated RL agent models were pre-trained offline for suspension structures with different numbers of components, forming a "suspension structure-controller parameter" mapping model library. The application phase is as follows... Figure 6 As shown, the corresponding model can be called for inference based on the number of components, thereby improving optimization efficiency.

[0103] The ISD suspension structure is a nine-element combination topology. Each three-element unit contains one damping element, one spring element, and one capacitance element. The system consists of three sets of three-element units, with two sets connected in series and then connected in parallel with the third set, ultimately forming a nine-element structure.

[0104] Active control employs the MPC method, implemented using the Simulink / MPC toolbox. Within each sampling period, the controller transforms the state feedback and reference tracking problem into a quadratic programming problem with linear constraints. This quadratic programming problem is automatically constructed by the MPC toolbox based on the configured model and constraints, and its built-in quadratic programming solver based on the effective set algorithm is invoked for online numerical solution to obtain the optimal control sequence and implement rolling optimization control.

[0105] The MPC toolbox requires configuration of matrices A, B, C, and D to describe the state-space model; a sampling time of 0.01 s to ensure a fast response to road surface stimuli; a prediction step size of 30 steps to define the time range for controller optimization; a control step size of 20 steps to limit the time range within which the optimization algorithm can freely adjust control actions; and a control force value range of [missing value]. The physical output limit of the active suspension actuator; the range of the control force change rate is... It is used to simulate the dynamic response speed limit of the actuator.

[0106] The dynamic equations for the 1 / 4 car dynamics model with active ISD suspension are:

[0107]

[0108]

[0109]

[0110] In the formula, For the sprung mass, in units ; For unsprung mass, in units ; For the equivalent spring stiffness of the tire, in units ; Damping coefficient, unit ; The spring stiffness coefficient, in units of ; The inertia coefficient, in units of ; As a power source, the unit ; The vertical displacement of the sprung mass, in units ; The vertical displacement of the intermediate connection point, in units. ; Vertical displacement of unsprung mass, in units ; For road surface excitation, unit .

[0111] Select the state vector of the active ISD suspension system:

[0112]

[0113] Select the system input for the active ISD suspension system:

[0114]

[0115] Select the system output of the active ISD suspension system:

[0116]

[0117] The state equation for the active ISD suspension is:

[0118]

[0119]

[0120] Substituting the dynamic equations into the state equations, we obtain the system matrix as follows:

[0121]

[0122]

[0123]

[0124]

[0125] In the formula, ,

[0126] ,

[0127] ,

[0128] ,

[0129] ,

[0130] ,

[0131] ,

[0132] ,

[0133] ,

[0134] ,

[0135] ,

[0136] ,

[0137] ,

[0138] ,

[0139] ,

[0140] ,

[0141] ,

[0142] ,

[0143] ,

[0144] ,

[0145] ,

[0146] ,

[0147] ,

[0148] ,

[0149] .

[0150] Output weight matrix Used for system output Each variable is weighted separately, and the input weight matrix is ​​used. It is used to limit the magnitude and variation of the active suspension control force.

[0151] This invention uses the filtered white noise method to describe the random excitation of a single-wheel road surface, and the formula is as follows:

[0152]

[0153] in, express The first derivative with respect to time, Indicates single-wheel road surface excitation. Indicates the reference space frequency. Indicates vehicle speed. This represents the road surface roughness coefficient. This represents ideal white noise.

[0154] The specific operational aspects of the GA–RL two-layer optimization framework:

[0155] Phase 1: Training Phase.

[0156] 1. System parameters and simulation environment initialization. Set the basic vehicle parameters and driving conditions, including: sprung mass, unsprung mass, tire equivalent spring stiffness, vehicle speed, and road roughness coefficient; based on these parameters, build a vehicle dynamics model with a nine-element ISD suspension and an active suspension MPC controller framework on the Simulink simulation platform.

[0157] 2. Initialization and Parameter Encoding of the Upper-Level Genetic Algorithm (GA). Define each individual in the genetic algorithm as a nine-dimensional real-valued vector, with each gene corresponding to the parameter value of an element. Set a parameter existence threshold (e.g., 100). When a gene value is below this threshold, it is determined that the corresponding element "does not exist" in the suspension structure, thereby achieving synchronous optimization of the suspension topology.

[0158] Genetic parameter settings: The population size of GA is set to 80, the maximum number of iterations is 30, the multi-point crossover rate is 0.7, the number of crossover points is 6, and the mutation rate is 0.4.

[0159] 3. Lower-level reinforcement learning (RL) initialization and controller parameter mapping. The five key parameters of the MPC controller ( , , , and ( ) serves as the action space for RL agents.

[0160] Training settings: The training rounds of the RL agent are set to 200, the initial exploration rounds are set to 256, and the experience replay buffer saves data starting from the 5000th step to ensure training stability.

[0161] 4. Set up a target category set based on the number of components. Before optimization begins, define a target category set based on the number of components, with values ​​ranging from 2 to 9. If a component's number of components belongs to the target category set, proceed to the next step; otherwise, skip that component in the current iteration and do not proceed to the next level of optimization.

[0162] 5. Two-layer optimization evaluation strategy. Initial evaluation: In the first 30% of iterations of the genetic algorithm, structural rationality is checked for all individuals in the population, and lower-level RL training is performed. Later elite evaluation: In subsequent iterations, an "elite + random" strategy is adopted. Only the top 20% of elite individuals and no more than 10% of random individuals are subjected to structural rationality checks and lower-level training. The remaining individuals use historical evaluation values, significantly improving optimization efficiency.

[0163] 6. Convergence Criteria Setting. Set the composite conditions for algorithm termination:

[0164] The fluctuation range of the optimal fitness value (fitness) over 10 consecutive generations is less than the threshold of 0.001;

[0165] The system output performance meets the following requirements: root mean square value of sprung mass acceleration. Root mean square value of suspension travel Root mean square value of tire dynamic deflection .

[0166] When all of the above conditions are met, the result is considered convergent.

[0167] 7. GA–RL double-layer optimized main loop.

[0168] 7.1: Upper-level GA population generation and structure selection. For each individual in each generation of the GA population, a structural rationality check is performed. If the combination of component parameters leads to model anomalies, the individual is discarded and assigned a very poor fitness value. For rational individuals, their nine-component parameters are passed to the lower-level RL module.

[0169] The initial population generation rule consists of two parts: one part is randomly generated within the parameter domain using a uniform distribution, and the other part is generated using a normal distribution centered on the current optimal solution. This allows the algorithm to converge towards the potential optimal region in the early stages, balancing search range coverage with rapid aggregation towards the optimal region.

[0170] The structural rationality test considers any of the following conditions to be unreasonable, and all others to be reasonable.

[0171] Case 1: c12, k12, and b12 are all absent, or c13, k13, and b13 are all absent (i.e., the damping coefficient, spring stiffness coefficient, and mass inertia coefficient of the first three-element unit are all 0, or the damping coefficient, spring stiffness coefficient, and mass inertia coefficient of the second three-element unit are all 0). In this case, the active ISD suspension system does not meet the dual-layer condition.

[0172] Case 2: C12, B12, C13, and B13 are all absent, while K12 and K13 are present (i.e., the damping coefficient and inertia coefficient of the first three-element unit are both 0, the damping coefficient and inertia coefficient of the second three-element unit are both 0, and the spring stiffness coefficients of both the first and second three-element units are not 0). In this case, the series connection of the active ISD suspension system only contains springs, which is of no practical significance.

[0173] Case 3: Neither b12 nor b13 exists (i.e., the inertia coefficients of both the first and second element units are 0). In this case, the denominator of the state equation is 0, which is unreasonable.

[0174] Case 4: k23 is nonexistent, and at least one of k12 and k13 is nonexistent (i.e., the spring stiffness coefficient of the third three-element unit is 0, and at least one of the spring stiffness coefficients of the first and second three-element units is 0). In this case, there is no continuous spring between the sprung mass and the unsprung mass, the support force is interrupted, and it cannot be guaranteed that the suspension will always have effective support capacity during operation.

[0175] 7.2: Lower-layer RL optimization. The RL agent receives suspension structure parameters from the upper layer and outputs a set of MPC controller parameters based on this structural environment.

[0176] 7.3: Performance Simulation. Substitute the combined structural and controller parameters into the Simulink model established in step 1 for simulation. Extract the root mean square values ​​of three key performance indicators from the simulation results: sprung mass acceleration... Suspension travel Tire dynamic deflection .

[0177] 7.4: Fitness and Reward Calculation. The fitness value (fitness) of the individual is calculated using the following formula, with the optimization objective being to minimize it:

[0178]

[0179] in , , These are the normalized weighting coefficients for each performance indicator.

[0180] To train an RL agent, the reward is defined as... A larger reward indicates that the controller parameters perform better under this structure.

[0181] 7.5: Experience Learning and Parameter Update. The state, action, reward, and next state data obtained from this simulation are stored in the RL experience playback buffer. The RL agent samples data from the buffer for training, updating its network weights to learn to generate better MPC controller parameters for a given suspension structure. The current optimal parameters are then updated. The value is used as the fitness of the corresponding GA individual and fed back to the outer genetic algorithm.

[0182] 7.6: Genetic Iteration. The outer genetic algorithm performs selection, crossover, and mutation operations based on the fitness of all individuals to generate a new generation of the population.

[0183] The selection process specifically includes:

[0184] First, the fitness value of all individuals in the current population is taken as the reciprocal and converted into a "base value of selection probability" that is proportional to the performance.

[0185] Next, calculate the sum of all individual baseline values, sumfitness, and divide each individual's baseline value by this sum to obtain the normalized selection probability sumf. Then, calculate the cumulative distribution cumsumf of these probabilities to prepare for roulette wheel selection.

[0186] Finally, a selection process is performed to determine the population size. Each selection generates one [population size]. The random number `pick`, uniformly distributed within the interval, is used to determine the index of the selected individual by finding the first position in the cumulative distribution `cumsumf` that is greater than or equal to `pick`. All the selected individual indices form an index array, which is used to replicate a new generation of the population from the current population, completing the selection process based on fitness ratios.

[0187] The multi-point intersection process specifically includes:

[0188] First, iterate through every position in the new population. Each time, randomly select two distinct individuals from the population to form a pair. For each pair, generate a... If a random number within the interval is less than or equal to the preset crossover probability pc, then a crossover operation is performed on the pair of individuals.

[0189] Secondly, if crossover is decided upon, six unique locations are randomly selected from all gene locations as crossover points. At each crossover point, the gene values ​​of the two parent individuals at that location are obtained. and .

[0190] Finally, generate one The random number alpha within the interval is used as the linear combination coefficient. The two parent gene values ​​are linearly interpolated and mixed, and the gene value of the first offspring at that position is updated to... The gene value of the second offspring at this position is updated to... Once this operation has been performed at all selected intersection points, the multi-point linear intersection of the pair of individuals is complete.

[0191] The mutation process specifically includes:

[0192] First, iterate through every individual in the new population. For each individual, generate a... A random number pick is generated within the interval. If the pick is less than or equal to the preset mutation probability, a mutation operation is performed on that individual. During mutation, a gene position pos is randomly selected from that individual as the mutation point.

[0193] Secondly, dynamic variable-length computation. Let the current gene value be... Its feasible value range is Calculate the distance from the current value to the lower limit. and the distance to the upper limit .

[0194] In the formula, Representing the The lower limit of the values ​​of each gene parameter Representing the The upper limit of the value of each gene parameter.

[0195] Finally, generate one again. Random numbers are picked within an interval to determine the direction of mutation. The calculation of the variable-length delta employs a non-uniform strategy that adaptively adjusts with the number of generations i, specifically:

[0196] like The mutation direction is upward, increasing the gene value, and the step size is... The new gene value .

[0197] like The mutation direction is downward, reducing the gene value, and the step size is... The new gene value This strategy results in a larger variable time length in the early stages of evolution to facilitate global exploration, and a smaller length in the later stages to facilitate fine-grained local searches. In the formula, This represents the maximum number of generations.

[0198] 7.7: Return to step 7.1 and continue iterating.

[0199] 8. Based on the component number classification set set in step 4, eight reinforcement learning agents corresponding to component numbers 2-9 can be obtained. All of them have been trained and matured through the two-layer optimization process and together constitute a "suspension structure-optimal controller parameter mapping model library".

[0200] Phase Two: Application Phase.

[0201] 9. Cooperative search of optimal structure and control parameters.

[0202] 9.1: Upper-level GA optimization. Maintain the same GA settings and parameter encoding method as in the first stage (set the maximum number of iterations to 150 generations), and start a new genetic algorithm optimization process with the goal of searching for the globally optimal nine-element structure parameters.

[0203] 9.2: Fast Lower-Level Evaluation. In this stage, lower-level optimization no longer involves time-consuming RL network training. For each individual to be evaluated generated by GA, a structural rationality check is performed, and the number of effective components is calculated. Based on the number of components, the corresponding RL model trained in the first stage is directly invoked. This model uses forward inference to output a set of optimal MPC controller parameters in real-time based on the current structural parameters.

[0204] 9.3: Performance simulation evaluation and iteration. Perform performance simulation (same as step 7.3), fitness calculation (same as step 7.4), and feed the fitness value back to the upper-level genetic algorithm to drive it to perform selection, crossover and mutation operations (i.e. the core logic of steps 7.5 and 7.6, but excluding the training and update steps of the RL network).

[0205] 9.4: Convergence Check. Check if the convergence conditions set in step 6 are met. If they are met, terminate the loop and output the current globally optimal nine-element suspension structure parameters and the five parameters of the MPC controller; otherwise, return to step 9.1 to continue iteration.

[0206] 10. Output and Application. The final output of the globally optimal nine-element suspension structure parameters and the five parameters of the MPC controller will be used as the final design scheme for the vehicle's ISD suspension system and applied to the suspension hardware configuration and active control system of the actual vehicle.

[0207] The technical effectiveness of the parameter design method for the active ISD suspension provided by the present invention will be verified through a specific embodiment below.

[0208] Select the sprung mass of a certain SUV model Unsprung mass Tire equivalent spring stiffness The damping coefficient of the passive SD suspension is The spring stiffness coefficient is .

[0209] Establish a Class C stochastic road surface model and take... , Vehicle speed The fitness value reached a stable convergence state after 40 generations, verifying the reliability of the optimization process.

[0210] After multiple iterations, the optimal suspension parameters are obtained as follows:

[0211]

[0212] The control parameters are:

[0213]

[0214] Set the vehicle to a constant speed The vehicle was driven on a Class C random road surface. Based on this condition, a whole-vehicle dynamics model was built in the MATLAB / Simulink environment and time-domain simulation was performed.

[0215] The root mean square value of sprung mass acceleration of active ISD suspension, passive ISD suspension and passive SD suspension Root mean square value of suspension travel and the root mean square value of tire dynamic deflection As shown in the table below.

[0216]

[0217] Comparative analysis of active ISD suspension and passive SD suspension shows that the root mean square value of sprung mass acceleration is optimized. Optimization of the root mean square value of suspension travel Optimization of the root mean square value of tire dynamic deflection .

[0218] The quantitative results above demonstrate that this invention achieves synergistic optimization in vehicle suspension system ride comfort, workspace safety, and handling stability. It not only significantly improves ride comfort but also ensures reliable suspension system operation and enhances tire-road grip. Furthermore, it provides a practical system-level technical solution for a comprehensive improvement in the overall driving and riding quality of the vehicle.

[0219] To better implement the parameter design method for the active ISD suspension in this embodiment of the invention, based on the parameter design method for the active ISD suspension, correspondingly, as follows: Figure 7 As shown, this embodiment of the invention also provides a parameter design device for an active ISD suspension. The parameter design device 700 for an active ISD suspension includes:

[0220] Module 701 is used to construct the vehicle dynamics model of the active ISD suspension system and the MPC controller framework of the active ISD suspension system based on the sprung mass, unsprung mass, tire equivalent spring stiffness, vehicle speed, and road roughness coefficient of the target vehicle. The active ISD suspension system is composed of a first three-element unit connected in series with a second three-element unit and then connected in parallel with a third three-element unit. Each three-element unit contains a damping element, a spring element, and an inertial capacitance element. The vehicle dynamics model is used to solve for the damping coefficient, spring stiffness coefficient, and inertial mass coefficient of each three-element unit. The MPC controller framework is used to solve for the input weights and output weights of the active ISD suspension system.

[0221] The determination module 702 is used to determine the damping coefficient, spring stiffness coefficient, and inertia coefficient of each three-element unit based on the GA algorithm and the vehicle dynamics model. Based on the determined damping coefficient, spring stiffness coefficient, inertia coefficient, MPC controller frame, and RL model of each three-element unit, the input weights and output weights of the active ISD suspension system are determined. The individual fitness of the GA algorithm is determined based on the sprung mass acceleration, suspension dynamic travel, and tire dynamic deflection of the active ISD suspension system. During training, the RL model uses multiple sets of sample damping coefficients, sample spring stiffness coefficients, and sample inertia coefficients determined by the GA algorithm as samples, the input weights and output weights of the active ISD suspension system as the action space, and the reciprocal of the individual fitness of the GA algorithm as the reward function.

[0222] The active ISD suspension parameter design device 700 provided in the above embodiments can realize the technical solutions described in the above active ISD suspension parameter design method embodiments. The specific implementation principles of each module or unit can be found in the corresponding content in the above active ISD suspension parameter design method embodiments, which will not be repeated here.

[0223] like Figure 8 As shown, the present invention also provides a parameter design device 800. The parameter design device 800 includes a processor 801, a memory 802, and a display 803. Figure 8 Only a portion of the components of the parametric design device 800 are shown; however, it should be understood that implementation of all shown components is not required, and more or fewer components may be implemented instead.

[0224] In some embodiments, processor 801 may be a central processing unit (CPU), microprocessor, or other data processing chip, used to run program code stored in memory 802 or process data, such as the parameter design method for active ISD suspension in this invention.

[0225] In some embodiments, processor 801 may be a single server or a group of servers. The server group may be centralized or distributed. In some embodiments, processor 801 may be local or remote. In some embodiments, processor 801 may be implemented on a cloud platform. In one embodiment, the cloud platform may include a private cloud, public cloud, hybrid cloud, community cloud, distributed cloud, internal cloud, multi-cloud, etc., or any combination thereof.

[0226] In some embodiments, memory 802 may be an internal storage unit of the parametric design device 800, such as a hard disk or memory of the parametric design device 800. In other embodiments, memory 802 may also be an external storage device of the parametric design device 800, such as a pluggable hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the parametric design device 800.

[0227] Furthermore, the memory 802 may include both internal storage units of the parameter design device 800 and external storage devices. The memory 802 is used to store the application software and various types of data installed on the parameter design device 800.

[0228] In some embodiments, display 803 may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an organic light-emitting diode (OLED) touchscreen. Display 803 is used to display information from the parametric design device 800 and to display a visual user interface. Components 801-803 of the parametric design device 800 communicate with each other via a system bus.

[0229] In one embodiment, when processor 801 executes the parameter design program for the active ISD suspension in memory 802, the following steps can be implemented:

[0230] Based on the sprung mass, unsprung mass, tire equivalent spring stiffness, vehicle speed, and road roughness coefficient of the target vehicle, a vehicle dynamics model and an MPC controller framework for the active ISD suspension system are constructed. The active ISD suspension system is composed of a first three-element unit connected in series with a second three-element unit and then connected in parallel with a third three-element unit. Each three-element unit contains a damping element, a spring element, and an inertial capacitance element. The vehicle dynamics model is used to solve for the damping coefficient, spring stiffness coefficient, and inertial mass coefficient of each three-element unit. The MPC controller framework is used to solve for the input weights and output weights of the active ISD suspension system.

[0231] The damping coefficient, spring stiffness coefficient, and inertia coefficient of each three-element unit are determined based on the GA algorithm and vehicle dynamics model. Based on the determined damping coefficient, spring stiffness coefficient, inertia coefficient, MPC controller frame, and RL model of each three-element unit, the input weights and output weights of the active ISD suspension system are determined. The individual fitness of the GA algorithm is determined based on the sprung mass acceleration, suspension dynamic travel, and tire dynamic deflection of the active ISD suspension system. During training, the RL model uses multiple sets of sample damping coefficients, sample spring stiffness coefficients, and sample inertia coefficients determined by the GA algorithm as samples, the input weights and output weights of the active ISD suspension system as the action space, and the reciprocal of the individual fitness of the GA algorithm as the reward function.

[0232] It should be understood that when the processor 801 executes the parameter design program for the active ISD suspension in the memory 802, in addition to the functions mentioned above, it can also perform other functions, as detailed in the description of the corresponding method embodiments above.

[0233] Furthermore, this embodiment of the invention does not specifically limit the type of the parameter design device 800 mentioned. The parameter design device 800 can be a portable electronic device such as a mobile phone, tablet computer, or personal digital assistant (PDA). Exemplary embodiments of portable electronic devices include, but are not limited to, portable electronic devices running iOS, Android, Microsoft, or other operating systems. The aforementioned portable electronic devices can also be other portable electronic devices, such as portable computers with touch-sensitive surfaces (e.g., touch panels). It should also be understood that in some other embodiments of the invention, the parameter design device 800 may not be a portable electronic device, but rather a desktop computer with a touch-sensitive surface (e.g., a touch panel).

[0234] Accordingly, this application also provides a computer-readable storage medium for storing a computer-readable program or instruction. When the program or instruction is executed by a processor, it can implement the steps or functions in the parameter design method for the active ISD suspension provided in the above-described method embodiments.

[0235] Those skilled in the art will understand that all or part of the processes of the methods described in the above embodiments can be implemented by a computer program instructing related hardware (such as a processor, controller, etc.), and the computer program can be stored in a computer-readable storage medium. The computer-readable storage medium may be a disk, optical disk, read-only memory, or random access memory, etc.

[0236] The above provides a detailed description of the parameter design method and device for the active ISD suspension provided by the present invention. Specific examples have been used to illustrate the principle and implementation of the present invention. The description of the above embodiments is only for the purpose of helping to understand the method and core idea of ​​the present invention. At the same time, for those skilled in the art, there will be changes in the specific implementation and application scope based on the idea of ​​the present invention. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A parameter design method for an active ISD suspension, characterized in that, include: Based on the sprung mass, unsprung mass, tire equivalent spring stiffness, vehicle speed, and road roughness coefficient of the target vehicle, a vehicle dynamics model and an MPC controller framework for the active ISD suspension system are constructed. The active ISD suspension system is composed of a first three-element unit connected in series with a second three-element unit and then connected in parallel with a third three-element unit. Each three-element unit contains a damping element, a spring element, and an inertial capacitance element. The vehicle dynamics model is used to solve for the damping coefficient, spring stiffness coefficient, and inertial mass coefficient of each three-element unit. The MPC controller framework is used to solve for the input weights and output weights of the active ISD suspension system. The damping coefficient, spring stiffness coefficient, and inertia coefficient of each three-element unit are determined based on the GA algorithm and vehicle dynamics model. Based on the determined damping coefficient, spring stiffness coefficient, inertia coefficient, MPC controller frame, and RL model of each three-element unit, the input weights and output weights of the active ISD suspension system are determined. The individual fitness of the GA algorithm is determined based on the sprung mass acceleration, suspension dynamic travel, and tire dynamic deflection of the active ISD suspension system. During training, the RL model uses multiple sets of sample damping coefficients, sample spring stiffness coefficients, and sample inertia coefficients determined by the GA algorithm as samples, the input weights and output weights of the active ISD suspension system as the action space, and the reciprocal of the individual fitness of the GA algorithm as the reward function.

2. The parameter design method for the active ISD suspension according to claim 1, characterized in that, The vehicle dynamics model is based on the following formula: in, Indicates unsprung mass. Indicates the sprung mass. This indicates the equivalent spring stiffness of the tire. Indicates power. Indicates single-wheel road surface excitation. Represents the vertical displacement of the unsprung mass. This represents the vertical velocity of the unsprung mass. This represents the vertical acceleration of the unsprung mass. This represents the vertical displacement of the intermediate connection point. This indicates the vertical velocity at the intermediate connection point. This represents the vertical acceleration at the intermediate connection point. Represents the vertical displacement of the sprung mass. Represents the vertical velocity of the sprung mass. Represents the vertical acceleration of the sprung mass. This represents the damping coefficient of the first three-element unit. This represents the spring stiffness coefficient of the first three-element unit. This represents the mass inertia coefficient of the first three-element unit. This represents the damping coefficient of the second and third element units. This represents the spring stiffness coefficient of the second and third element units. This represents the mass inertia coefficient of the second and third element units. This represents the damping coefficient of the third element unit. This represents the spring stiffness coefficient of the third element unit. This represents the mass inertia coefficient of the third element unit.

3. The parameter design method for the active ISD suspension according to claim 2, characterized in that, The single-wheel road surface excitation is determined based on the following formula: in, express The first derivative with respect to time, Indicates single-wheel road surface excitation. Indicates the reference space frequency. Indicates vehicle speed. This represents the road surface roughness coefficient. This represents ideal white noise.

4. The parameter design method for the active ISD suspension according to claim 1, characterized in that, The MPC controller framework is represented by the following formula: in, This represents the state vector of the active ISD suspension system. Indicates the input of the active ISD suspension system. This indicates the output of the active ISD suspension system. express The first derivative with respect to time, Indicates power. Indicates single-wheel road surface excitation. Represents the vertical displacement of the unsprung mass. This represents the vertical velocity of the unsprung mass. This represents the vertical displacement of the intermediate connection point. This indicates the vertical velocity at the intermediate connection point. Represents the vertical displacement of the sprung mass. Represents the vertical velocity of the sprung mass. , , , This is the system matrix for the MPC controller framework.

5. The parameter design method for the active ISD suspension according to claim 1, characterized in that, The individual fitness of the GA algorithm is determined based on the following formula: in, Indicates individual fitness. Indicates the acceleration of the sprung mass. Indicates the suspension travel. Indicates tire deflection. , , These are the preset normalized weighting coefficients.

6. The parameter design method for the active ISD suspension according to claim 1, characterized in that, The convergence conditions of the GA algorithm include: The fitness values ​​of the GA population for 10 consecutive generations fluctuate within a range less than the fluctuation threshold, and the root mean square value of the sprung mass acceleration is less than or equal to the first threshold, the root mean square value of the suspension dynamic travel is less than or equal to the second threshold, and the root mean square value of the tire dynamic deflection is less than or equal to the third threshold.

7. The parameter design method for the active ISD suspension according to claim 1, characterized in that, The method further includes: Perform a structural rationality check on each individual in the GA population of each generation of the GA algorithm, and discard individuals that fail the structural rationality check; If a target individual meets at least one of the following conditions, it is determined that the target individual failed the structural rationality test. The target individual can be any individual in any generation of the GA population: The damping coefficient, spring stiffness coefficient, and mass inertia coefficient of the first three-element unit are all 0, or the damping coefficient, spring stiffness coefficient, and mass inertia coefficient of the second three-element unit are all 0. The damping coefficient and inertia coefficient of the first three-element unit are both 0, the damping coefficient and inertia coefficient of the second three-element unit are both 0, and the spring stiffness coefficient of the first three-element unit and the spring stiffness coefficient of the second three-element unit are both not 0. The mass inertia coefficients of both the first and second three-element units are 0. The spring stiffness coefficient of the third three-element unit is 0, and at least one of the spring stiffness coefficients of the first three-element unit and the second three-element unit is 0.

8. A parameter design device for an active ISD suspension, characterized in that, include: The module is used to construct the vehicle dynamics model and MPC controller framework of the active ISD suspension system based on the sprung mass, unsprung mass, tire equivalent spring stiffness, vehicle speed, and road roughness coefficient of the target vehicle. The active ISD suspension system is composed of a first three-element unit connected in series with a second three-element unit and then connected in parallel with a third three-element unit. Each three-element unit contains a damping element, a spring element, and an inertial capacitance element. The vehicle dynamics model is used to solve for the damping coefficient, spring stiffness coefficient, and inertial mass coefficient of each three-element unit. The MPC controller framework is used to solve for the input weights and output weights of the active ISD suspension system. The determination module is used to determine the damping coefficient, spring stiffness coefficient, and inertia coefficient of each three-element unit based on the GA algorithm and vehicle dynamics model. Based on the determined damping coefficient, spring stiffness coefficient, inertia coefficient, MPC controller frame, and RL model of each three-element unit, the input weights and output weights of the active ISD suspension system are determined. The individual fitness of the GA algorithm is determined based on the sprung mass acceleration, suspension dynamic travel, and tire dynamic deflection of the active ISD suspension system. During training, the RL model uses multiple sets of sample damping coefficients, sample spring stiffness coefficients, and sample inertia coefficients determined by the GA algorithm as samples, the input weights and output weights of the active ISD suspension system as the action space, and the reciprocal of the individual fitness of the GA algorithm as the reward function.

9. A parameter design device, characterized in that, Including memory and processor, among which, The memory is used to store programs; The processor, coupled to the memory, is used to execute the program stored in the memory to implement the steps in the parameter design method for the active ISD suspension according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, Used to store computer-readable programs or instructions, which, when executed by a processor, can implement the steps in the parameter design method of the active ISD suspension as described in any one of claims 1 to 7.

Images