Vehicle Multi-model Hybrid Low Speed Control Based On Generalized Predictive Control

Vehicle is a double-edged sword as scientific and technological achievements,which brings convenience to people's travel,but also caused many social problems.With the increase of vehicle ownership,traffic jams and safety,noise pollution and many other crucial issues have become gradually prominent.In recent years,the birth and development of Internet plus and artificial intelligence technology makes the human pursuit of more safe,efficient and convenient life,therefore,vehicle with Advanced Driver Assistance System emerged.As a sub function of ADAS,the adaptive cruise control system guarantees safe driving,and further improves the driving comfort and economy.Firstly,the low speed adaptive cruise control system functional requirements are defined in this paper.The system overall structure is divided into several modules according to their functions,and these functions and the key technologies requires for these functions are characterized.Then,the vehicle dynamic characteristics at low speed are analyzed,and a simplified vehicle longitudinal dynamic inverse model is obtained.The switching strategy of the driving and braking control based on the inverse model formula is implemented and can effectively avoid frequent switching.According to the engine nonlinear characteristic,large scale linearization method is used to compensate partial-nonlinear of vehicle,and the CARMA(controlled autoregressive moving average model)is adopted to describe the approximate liner systems.After system model polynomial orders are defined through structural identification,the CARMA model's parameters that express the real vehicle's characters are got based on the Least Square Method.Several driving parameter models and single braking parameter model are established through analyzing comparison between response of transfer function and actual system.Based on the above works,a multi-model hybrid low speed controller is established by using generalized predictive control algorithm.The controller adopt model to predict multi-step system response and the optimal control quantity of the current state of the system is calculated by the target function including weighted value of the actual acceleration and the expected acceleration error and the weighted value of the control increment.In order to verify the feasibility and effectiveness of the proposed method,the traditional PID controller is built to compare with the proposed method for performance indexes including response time,overshoot and steady-state error.The simulation results show that the response of the actual acceleration to the target acceleration is stable,and the oscillation and divergence of actual acceleration do not appear.The low controller combining multi parameter models and Generalized Predictive Control algorithm can improve vehicle control performance at low speed.When the vehicle internal parameter changing or small disturbance of the external environment appearing,the change of the internal parameters of the vehicle and the small change of the external environment,the controller can control the vehicle to return to a stable state quickly.