In view of an optimization of the tracking accuracy and tracking time of Maximum Power Point Tracking (MPPT) technology, an improved Adaptive Particle Swarm Optimization (APSO) algorithm has thus been proposed. An optimization can be achieved of the traditional PSO algorithm by introducing adaptive inertia weights and nonlinear learning factors so as to accelerate MPPT tracking in the global optimization-local optimization-global optimization state. Subsequently, a photovoltaic power generation system is to be established for an simulation and verification of the adaptive particle swarm optimization algorithm. Experimental results indicate that compared to traditional PSO algorithms, the improved APSO algorithm is characterized with a higher tracking accuracy and faster convergence speed. Under constant and variable temperatures in an unobstructed environment (STC), the convergence speed has increased by 30.6% and 39.2% respectively, while the convergence speed under constant and variable temperatures in a partial occlusion (PSC) has increased by 54.0% and 53.7%, showing a performance superiority of the improved APSO algorithm in PSC environments. Furthermore, the PSO algorithm exhibits oscillations in the duty cycle after stabilizing the maximum power, while the APSO algorithm maintains a stable duty cycle, thereby improving the overall stability of the system.