Abstract: High precision parameters are the key for permanent magnet synchronous motor to realize high performance control. To overcome the shortages of slow speed and low identification accuracy in traditional identification methods, a novel teaching-learning-based optimization algorithm with Levy flight is proposed to identify the PMSM parameters. The algorithm introduces adaptive teaching factor and self-learning strategy to improve the convergence speed. As for learning phase, a Levy flight stochastic process is introduced to improve the optimization strategy so that the algorithm can enhance the ability to keep the balance between exploration and exploitation. The simulation results show that the novel algorithm can accurately identify the stator resistance, d-axis, q-axis inductance and the rotor linkage with better convergence and reliability.

Key words: permanent magnet synchronous motor, parameter identification, Levy flight, adaptive teaching factor, self-learning strategy