Abstract: High accuracy identification of parameters in permanent magnet synchronous motor (PMSM) is the basis of controller design. In order to overcome the shortages of traditional identification methods such as slow speed and low identification accuracy, an improved teaching-learning-based optimization algorithm (ITLBO) was proposed to identify the permanent magnet synchronous motor parameters. In the teaching phrase, tutorial teaching mechanism was introduced to strengthen teacher's capacity and improved the convergence rate of algorithm, in the learning phrase, the course stepwise learning was used to improve learners' learning efficiency. Besides, opposition-based-learning was introduced for small probability mutation, which enhanced the possibility out of local optima. The simulation result shows that the proposed algorithm has better convergence and reliability in simultaneous identification of the stator resistance, d-axis and q-axis inductance and the rotor linkage compared with teaching-learning-based optimization and particle swarm optimization.

Key words: permanent magnet synchronous motor, parameter identification, teaching-learning-based optimization, tutorial teaching, opposition-based-learning, particle swarm optimization