Research On Optimization Control Of Vehicle Active Suspension Based On Long Short-Term Memory Neural Network

Recently,intelligent control of vehicle suspension system has gradually become the main research direction of intelligent by-wire chassis.The main function of suspension system is to mitigate vibration from uneven road surfaces and ensure the ride comfort and handling stability of the vehicle.Active suspension system(ASS)has become a research hotspot of new technology of vehicle chassis in recent years,because ASS can ensure ride comfort and safety of vehicle by self-adjusting control force according to the current environment and vehicle conditions.The research of ASS with excellent control strategy is an important technical content in the development of wire chassis system.Due to the deficiency of adaptive ability of the traditional control method,the controller can not always be in the optimal state to exhibit the best control effect.Therefore,the control method of ASS is further researched in this paper to achieve the goal of improving ride comfort of vehicle and improve the adaptive ability of the control algorithm.The main work is as follows:(1)The 2-DOF dynamic model of the quarter vehicle ASS is established according to the actual ASS,and the dynamics equations of the ASS can be formulated.Then,the bump road and random road are chosen as the input of ASS.Finally,the performance evaluation indexes of ASS are chosen according to the requirements of vehicle ride comfort.(2)Based on the dynamic model of the vehicle ASS,an enhanced vibration control method is presented for ASS by combining back propagation neural network(BPNN)and particle swarm optimization(PSO).The proposed BPNN is utilized to make the PID controller perform better.Meanwhile,the improved PSO is presented to obtain the weights and thresholds of the BPNN controller,which can improve the convergence rate and accuracy of the designed control system and avoid the designed BPNN falling into the local optimum.Besides,the Lyapunov stability theory is utilized to analyze the stability of the ASS with the proposed controller.Finally,a simulative investigation in MATLAB/Simulink is performed to demonstrate the effectiveness of the proposed controller.(3)Based on the idea of deep learning,the long short-term memory network(LSTM)control strategy of ASS based on dynamic event triggered(DET)mechanism is designed.First,a quarter-vehicle ASS model with input dead zone and saturation is constructed.Then,an appropriate DET controller is presented to solve the communication tension question.Meanwhile,the LSTM controller is used to make the vertical acceleration of the ASS close to zero and thus improve the ride comfort of vehicle.Besides,the radial basis function neural network is applied for the designed vibration controller to compensate the dead zone and saturation.Moreover,the overall system stability is certified with the help of Lyapunov theorem,and the Lyapunov theorem is applied to eliminate the Zeno phenomenon.Finally,the presented control method is proved to be effective by different road excitation.(4)According to the perdormance requirements of intelligent control of ASS,the software framework of ASS intelligent control performance analysis is designed.The intelligent control performance analysis software of vehicle ASS is designed by using MATLAB APP Designer,and the software is shown and explained in this paper.The performance analysis software is simple to operate,which has strong practicability.