The 2 640 MW large scale centralized grid-connected photovoltaic power station in Karamay，Xinjiang，is taken as an example in this paper，and an intelligent design method for photovoltaic power stations is proposed，based on the Particle Swarm Optimization （PSO） algorithm. This method uses photovoltaic module type selection，installation angle，array spacing and other design parameters as optimization variables，with annual power generation maximization as the objective function and investment costs as constraints，to construct a multi-objective optimization model. Power generation simulation is performed using the PVsyst software combined with MATLAB programming to implement PSO algorithm optimization. Simulation results show that compared with traditional design methods，the intelligent optimization method proposed in this paper can increase the annual power generation of the 2 640 MWphotovoltaic power stations by 3.2% to 41.26×10⁸ kW·h，and shorten the investment payback period by 0.5 years. After completion，the project can reduce CO₂ emissions by approximately 7.176×10⁷ t per year，with significant economic and social benefits. The proposed method provides new ideas for intelligent design of large scale photovoltaic power stations.