In the traditional maximum power point tracking (MMPT) algorithm, the multi-peak phenomenon of the P-U characteristic curve occurs when the local shadow appears in the photovoltaic array, which leads to a failure of the true maximum power point tracking, thus reducing the system output rate. Particle swarm optimization (PSO) works well in global search. PSO is used in MPPT, but it has some disadvantages in convergence speed and precision. In order to improve the tracking accuracy and convergence speed of the PSO Algorithm, a method of combining the nonlinear control theory strategy with the PSO algorithm is proposed, and Matlab/Simulink is used to do the simulation to verify its feasibility in this paper. The simulation results show that the improved PSO algorithm can track the maximum power point rapidly, stably and accurately under the condition of no shadow and environmental changes, which improves the power generation efficiency of the photovoltaic system.