Monitoring the permanent magnet temperature of permanent magnet motors was critical to the life of the motor, because the excessive permanent magnet temperature increased the risk of irreversible demagnetization. A method based on particle swarm optimization (PSO) algorithm for permanent magnet motor thermal network temperature identification was proposed to monitor the temperature of permanent magnets. In this method, the thermal network model of the permanent magnet motor was established, and the PSO algorithm was combined with the temperature data obtained by the motor temperature rise experiment to identify the main thermal parameters of the model. This method was used to identify the temperature online, and the identification process could be quickly converged with good accuracy. Finally, the feasibility and accuracy of the method were verified by comparing the temperature of the simulation with the experimental data.