To achieve low-cost operation of the data center park, a multi-energy system for the fully lake water natural cooling data center has been constructed. In view of the volatility of renewable energy power generation and the difference in peak-valley electricity prices, a fine-grained regulation strategy for load migration and storage/discharge optimized by particle swarm optimization based on specific spatio-temporal rule control was proposed. This strategy precisely matches the delay-tolerant computing power load with the output and electricity price periods of renewable energy, reducing configuration capacity redundancy and green electricity loss. The capacity configuration of energy storage equipment has decreased by 23.1%. The annual operating electricity cost is saved by 19.7%, and the green electricity loss rate in summer and the transitional season is reduced by 1.5% and 0.4% respectively. The power usage effectiveness (PUE) decreased from 1.121 to 1.115. The strategy provides a reference for the optimization of multi-energy and load scheduling.