When applying forgetting factor recursive least squares for permanent magnet synchronous motor (PMSM) parameter identification, relying on empirical fixed values for the forgetting factors often leads to reduced accuracy and poor convergence. To address this issue, this paper proposes an optimized strategy integrating whale optimization algorithm and artificial fish swarm algorithm to search for optimal forgetting factors. First, leveraging the complementary characteristics of both algorithms, a fusion algorithm with high convergence precision and robust performance is proposed. Subsequently, AWFSA is employed to optimize the objective function tailored for PMSM parameter identification. The optimized forgetting factors are integrated into the parameter identification module for online estimation and the smoothing transition module feeds back to the observer, thereby further enhancing rotor speed position estimation accuracy. Finally, simulations and bench tests demonstrate that the proposed method improves PMSM parameter identification accuracy, and enhances the observation accuracy of PMSM rotor speed position observers.