Accurate Torque Modeling for Spherical Reluctance Motor Based on BPNN and Fourier Series

In order to solve the problem that the multicoil torque of spherical reluctance motor cannot be obtained by simply superposing multiple single-coil torques, this article proposes an accurate torque modeling method based on back propagation neural network (BPNN) and Fourier series. The BPNN is adopted to compensate the torque superposition errors at different excitation currents and motion angles. The Fourier series is used to fit the torque curve generated by exciting single coil. Finally, the accurate torque model can be established by combining the torque superposition error compensation model and the Fourier series expressions. The results show that the proposed torque model can overcome the torque superposition error caused by simple superposition, and has good universality and high accuracy.