Research On Vehicle Chassis Coordinated Control Method Based On Multi-agent Theory

With the highly development of human society,civilization of science and technology,and the continuous improvement of the pace of life,the era of vehicles as a means of transport gradually coming.The number of car ownership is increasing every year,and people's requirements for the operation of the car driving and comfort requirements are also rising.The coordination control of vehicle chassis has became one of the most popular research topics,because it is an important factor that affects the vehicle comfort,ride,safety and other important factors.It was simulated in MATLAB/Simulink environment,based on the vehicle dynamics theory and combined with the relevant data collected by the vehicle through multi sensor acquisition.The mathematical model includes the brake agent model,the steering agent model and the 7 degree of freedom suspension agent model.Then,the reinforcement learning algorithm based on RBF neural network is applied to the coordinated control of the chassis multi-agent system.It makes use of the cereber model articulation controller(CMAC)algorithm to reduce and generalize the continuous large amount of data provided by the suspension agent,steering agent and brake agent.Then,the enhancement signal is determined by the characteristics of the chassis of each agent.The learning network and the evaluation network of the reinforcement learning algorithm are trained to utilize the RBF neural network in order to optimize the performance of the local agent and improve the comfort,stability and safety of the vehicle.Secondly,the weight of performance indicators in reinforcement learning is determined by fuzzy control method.The fuzzy controller of performance indicators weights of are designed,according to the experience rules.The empirical rules are obtained from the data under different operating conditions.The design of the vehicle chassis multi-agent used enhanced learning control strategy was completed.Finally,in the MATLAB/Simulink simulation environment,the design of the Multi-Agent Reinforcement learning algorithm is used in the test bench.The test bench is xPC Target real-time simulation system hardware in the loop test bench.The simulation results are analyzed in detail in the braking conditions of the car,the moment of turning to the driving conditions,and the steering braking complex driving conditions.It validates the effectiveness of the multi-agent coordination control algorithm proposed in this paper to improve vehicle comfort,safety and ride comfort.