| PID control algorithm which is widely used in most of current diesel electronic speed governor is simple and practical, but it has inevitable disadvantages that its control performance becomes bad when the controlled object has the characteristics of nonlinear and time-varying. In the process of the application of traditional PID control strategy, to calibrate the control parameters of PID aiming at a kind of diesel,a large amount of experiments have to be done and the resources consumption is large.To improve the performance of PID control algorithm, an adaptive control algorithm which is based on the idea that combines BP neural network theory with the traditional PID control strategy is put forward in this paper. The algorithm can real-timely optimize the control parameters of PID. By using the self-learning ability of BP neural network, we can achieve the purpose of optimizing the performance of diesel speed governor online.According to the working principle of the diesel engine, a model of diesel speed control system is developed on the platform of Matlab/Simulink, providing a good simulation environment of new diesel speed control algorithm. Simulation results show that the adaptive speed control algorithm can fulfill the task that optimize the control parameters automatically and the performance of the adaptive PID control algorithm is much better than that of the traditional PID control algorithm.To further study the application of this new control algorithm in diesel speed governor system, the STM32 is used as the core microcontroller to develop a diesel speed governor with hardware and software designing. Experiments have been done on diesel test bench. The results show that the self-adaptive engine speed governor can complete the task of speed governor taking account of the speed stability of diesel both at high speed and low speed. The performance of neural network adaptive PID algorithm which performance can meet the index of GB/T3475-2008 at both transient speed regulation and speed recovery time is much better than the traditional PID algorithm. |
