Neural Network-based MTPA Control Considering Temperature Variation in IPMSMs

This paper proposes an artificial neural network (ANN) based maximum torque per ampere (MTPA) control algorithm for interior permanent magnet synchronous motors (IPMSMs), considering temperature variations. As the temperature rises, the flux of the permanent magnets inside the rotor decreases, due to a reduction in residual flux density. Also, the d-q axis flux varies according to d-q axis current due to flux saturation as well as the variation of magnet flux. These variations in the d-q axis flux are nonlinear and, the changes of MTPA parts and output torque. The d-q axis flux variation due to the temperature change can be identified with the change of d-axis flux and d-q axis currents. The nonlinear relationship between the variation of d-axis flux and the MTPA point can be represented using ANN. The proposed algorithm is validated through experiments using an 11 kW IPMSM.