Research on torque model based on pretreatment method for switched reluctance motor

Owing to high nonlinearity of the electromagnetic characteristic caused by double-salient structure and high magnetic saturation, it is hard to establish an accurate model using formulas for switched reluctance motor (SRM). This paper establishes a torque model based on good nonlinear mapping ability of BP neural network. And for further optimization, a third input variable reflecting the system nonlinear characteristics is added to establish a BP neural network for SRM using pretreatment method. Comparing to traditional torque model, the proposed model has a small-scale network structure and higher accuracy, which is propitious to achieve real-time online control for SRM driving system.