Under dynamic illumination, the peak power of the photovoltaic array varies with the illumination. Traditional MPPT control methods only consider a single peak, resulting in low switching efficiency and poor tracking stability between different power peaks. To address this issue, a distributed photovoltaic multi peak MPPT control method based on an improved PSO algorithm is proposed. Firstly, analyze the U-I characteristics of the system and construct a simplified equation to establish a multi peak MPPT control model; Derive and simplify the U-I characteristic curve equation of distributed photovoltaic power generation system, and construct a control model with the goal of maximizing power. Using particle swarm optimization algorithm to solve the model, optimizing the initial position of particles through Tent mapping, adjusting the inertia weight through adaptive algorithm, and improving the algorithm by replacing fixed parameters with asynchronous learning factors. Experimental verification shows that this method has significant advantages in U-I characteristic analysis accuracy (average absolute error 0.08A, relative error 0.12%), dynamic tracking efficiency (up to 98.6%, 6.2 percentage points higher than traditional methods), response speed (shortened from 0.25s to 0.12s, 52% faster), and adaptability to complex environments (load satisfaction rate increased from 78% to 95% under 50% local shadow coverage), providing a reliable solution for efficient operation of photovoltaic systems under dual carbon targets.