Adaptive PID control and on-line identification for switched reluctance motors based on BP neural network

Switched reluctance motors (SRM) are favored in a lot of industrial applications because of many special characteristics they possess. But one of the disadvantages for SRMs is that they´re difficult to control, as a result of their nonlinear construction. This paper presents a new control solution: adaptive PID control and on-line identification based on BP neural network (BPNN). This method takes advantage of BPNN, which has strong self-learning and adaptive capabilities, to adjust the three parameters K_P , K_I , KD of the PID controller, and builds another three-layer BPNN as an on-line identification structure for the SRM to improve the new controller´s accuracy. With the proposed method, satisfying response speed and precision as well as good robust and stable performance has been obtained by experiments based on DSP.