Multi-Coil Torque Superposition Error Analysis and Estimation for Spherical Reluctance Motor

The nonlinear electromagnetic characteristics of spherical reluctance motor (SPRM) make the torque analysis quite difficult. This paper proposes a method to estimate the error of torque direct superposition based on backpropagation neural network (BPNN) with pretreatment. The finite element analysis (FEA) is used to obtain the single coil torque and the synthesized torque data of adjacent coils on the same layer. The spin angle, pitch angle, excitation current and pretreated data are used as the inputs, and the error of the torque direct superposition is used as the output for BPNN model. The simulation results show that this method can accurately predict the error of the torque superposition of adjacent coils.