In view of the low rank of the multi-parameter identification mathematical model of permanent magnet synchronous motor (PMSM), the motor parameters are divided into two types: fast change and slow change. The slow identification model and fast identification model are established by combining the slow parameter entry and the steady-state information of the motor, and a step-by-step iterative online identification method is proposed. Based on the steady-state voltage equation of the motor in the rotating coordinate system, the slowly changing parameters is fixed in time and step by step. The forgetting factor recursive least square algorithm can identify the three electrical parameters of the surface permanent magnet synchronous motor (SPMSM). It has small calculation pressure and fast identification speed, which is suitable for engineering application. Experiments are performed based on DSP controller and dynamometer test platform. The experimental results prove that the identification method is feasibility.