| Popis: |
The model of a switched reluctance motor (SRM) directly influences it’s performance and control effect. However, the doubly salient structure , switching power supply and seriously saturated magnetic circuit, make it difficult to accurately model. Aiming to solve this problem, this paper presents a new simulation method based on BP neural network to establish the current and torque models of a SRM. Using a great deal of current, flux linkage, torque numerical solution data obtained by a finite element software as the sample data, a kind of BP neural network is trained offline, and all the expected data sets, which form a table, are outputted. By looking-up the table the nonlinear models of flux linkage and torque of the SRM are obtained. On the basis of the nonlinear models, the current closed-loop control is formed by using the torque sharing function and the fuzzy switching angles to reduce the torque ripple. The simulation results show that the proposed nonlinear models not only have a fast calculation speed, but also are high precision and can meet the requirements of real-time control. The proposed control method can effectively reduce the torque ripple of the SRM. |
