Abstract: To address the common problems of traditional park microgrids, including low renewable energy consumption, poor multi-energy coordination, and high operating costs, this study draws on the practical experience of a science and technology park microgrid in Shanghai. It constructs an integrated source-network-load-storage multi-energy complementation system and establishes a closed-loop scheduling framework covering forecasting, optimization, execution, and correction. This study adopts a two-stage scheduling strategy that combines intraday rolling optimization with real-time correction. It also uses the analytic hierarchy process and entropy weight method to determine multi-objective weights. The results show that the proportion of self-generated and self-consumed photovoltaic electricity increases from 41.2% to 73.5%, annual operating cost decreases by 34.6%, and carbon emissions decrease by 45.2%. This paper reviews the key technical stages, problem-solving paths, and practical experience in the implementation of the microgrid project in Science and Technology Park A. It provides a replicable practical solution for the optimal scheduling of similar park microgrids.