针对标准PSO算法在优化过程后期易过早收敛问题，提出一种改进PSO算法，并对电力推进无人潜艇感应推进电机进行参数辨识.改进PSO算法通过对学习因子的调节，使粒子在优化过程初期具有较强的全局搜索能力，在优化过程后期快速收敛于全局最优解.改进PSO算法以感应推进电机的dq轴实际输出电流、电压作为参数辨识系统的输入，将电机的实际输出电流和电气模型的观测电流之间的差方和作为目标函数.通过实验，将改进粒子群优化算法、标准粒子群优化算法和遗传算法所辨识出的参数进行对比，结果表明，改进PSO算法可以获得更准确的感应推进电机辨识参数.

Aiming at the problem of standard PSO algorithm easy to converge at the end stage of optimization process, the paper proposed an advanced particle swarm optimization (PSO) for parameter estimation of marine induction propulsion motor in electric propulsion unmanned submarine. The advanced PSO modified the learning coefficients so as to improve the global search capability in the early stage of the optimization process, and then converged particles to the global optimum at the end stage. The advanced PSO algorithm used the difference between the measurements of the dqaxis currents of induction propulsion motor and the estimation currents of electrical model as the objective function. Estimated parameters comparison of induction propulsion motor among the advanced PSO algorithm, the standard PSO algorithm and genetic algorithm show that the advanced PSO algorithm can get more accurate estimated parameters of induction propulsion motor.