A microgrid is a small-scale power grid system composed of multiple distributed power sources and loads, characterized by energy interconnection, intelligence, high efficiency, and energy conservation. It serves as a critical component of future smart grids. However, the random fluctuations in electric vehicle (EV) charging loads pose significant challenges to the optimal scheduling of microgrid operations. To achieve efficient and stable microgrid operation, this paper proposes a microgrid cluster optimal scheduling strategy based on an Improved Particle Swarm Optimization (IPSO) algorithm. Building upon the traditional Particle Swarm Optimization (PSO), a normal distribution attenuation strategy is introduced in the early optimization stage, and a real-time learning factor update strategy is implemented in the later stage. These enhancements enable the algorithm to escape local optima effectively and address issues such as low solution accuracy. Comparative simulations with PSO, Sparrow Search Algorithm (SSA), and Grey Wolf Optimization (GWO) demonstrate that IPSO achieves higher optimization precision and effectively reduces the economic costs of microgrid cluster systems.