At present, Many longitudinal speed tracking controllers of smart cars adopt the PID control algorithm. The PID controller’s advantages of simple structure and easy reality are widely applied in industry. With the advantages of simple structure and easy reality, the controller is widely used in industry. But like most traditional controls, PID controllers do not apply to nonlinear-and time-varying systems, and can only be used for a fixed control object. The parameters of PID controllers need to be readjusted when the control objects are changed. In order to solve the problem, the paper uses a multi-layer feed forward neural network to adjust the PID parameters according to the different control objects, and makes it adaptive for a time-varying nonlinear longitudinal speed control system. The designed controller is verified by the matlab and carsim soft. It shows that the controller keeps the vehicle tracking error only within 0.05-0.06m/s, and the control system can respond quickly and achieve stability within 2 seconds when a signal mutation occurs. Thus, the effectiveness of the designed algorithm is demonstrated.

Full text: IJEIR_2914_FINAL.pdf