Research On Control Strategy Of MRD Semi-Active Suspension Based On Neural Network

Magneto-Rheological Damper(MRD)as an important execution element for semi-active suspension control systems,it has the advantages of low energy consumption,fast response speed,and wide range of control.Wide research in mechanical structures have been widely studied.Based on the neural network,this article gives the study of MRD semi-active suspension control strategies with damping and current as the direct control to improve the performance of the semi-active suspension of the car.The 1/4 suspension model of the vehicle is taken as the research object,the MRD mechanical model and the dynamic equation of the suspension system are established,and the relevant transmission functions of the suspension system evaluation index are derived.The passive and semi-active suspension simulation models built on random roads and impact roads under the Matlab/Simulink software provided a good condition for the study of the MRD semi-active control strategy.The test obtained the data of the MRD input current,the damping itinerary,the speed of the piston,and the damping force.Based on the test data,the PSO algorithm was used to correct the parameter recognition of the modified Bouc-Wen model to meet the MRD mechanical characteristics obtained by the test.In order to obtain the reverse characteristics of the MRD output damping force and the input current,considering the complexity of modifying the BOUC-WEN model structure,it is not suitable for solving its inverse model,and the method of using neural networks to obtain MRD inverse models.In the Matlab/Simulink environment,the fixing Bouc-Wen model and MRD inverse model are simulated.As a result,the fixing Bouc-Wen model and MRD inverse model can accurately describe the positive and inverse characteristics of MRD.Based on the MRD neural network inverse model,a 1/4 semi-active control suspension model with damping force as the control quantity is established.the control algorithms for the damping force of the sky-hook and the ground-hook canopy are weighted and optimized,and the results are obtained to balance the comfort and handling stability of the vehicle.In the Matlab/simulink environment,the passive suspension,the sky-hook control suspension,the ground-hook control suspension and the sky-hook and ground hook hybrid damping control suspension are simulated and analyzed.The comfort of the vehicle of the sky-hook and ground-hook hybrid damping control is lower than that of the sky-hook damping control,but it can restrain the deterioration of the vehicle handling by the sky-hook damping control.Compared with the sky-hook damping control suspension,the sky-hook and ground-hook hybrid damping control has a significant improvement effect on the comfort of the vehicle,and the improvement of the vehicle handling is not much different.In order to verify whether the mixed damping control strategy of the MRD semi-active suspension sky-hook and groundhook hybrid damping control has the optimal vibration reduction effect,it is compared with the optimal damping force control strategy of full state feedback.perform better.A 1/4 semi-active suspension model of the current as the control volume is established.Through the multi-layer BP neural network,the fuzzy PID controller’s three parameters of the three parameters are conducted offline.In the Matlab/Simulink environment,the passive suspension,the fuzzy PID control suspension,and the fuzzy-neurological PID control suspension are simulated and analyzed,the results show that the performance of the fuzzyneural PID semi-active control suspension on random roads and impact road surfaces is better than passive suspension and fuzzy PID semi-active control suspension.