The escalating frequency of extreme climatic events, particularly torrential rainfall, has acquired enhanced importance in spatial risk zoning for power grid infrastructure. In the study, we employ the Bayesian Structural Equation Model (BSEM) to realize risk zoning for power grids vulnerable to rainstorm disasters. Within the BSEM framework, a three-tiered risk assessment index system is developed, integrating meteorological, environmental, and historical power outage data. In contrast to conventional risk zoning methodologies, the proposed framework can effectively capture intricate causal interdependencies between diverse risk factors. By leveraging the constructed BSEM framework, we evaluate power grid risk indices (η) at sampled locations across Anhui Province, China, under rainstorm disaster scenarios. Then, the risk distribution of power grid in Anhui Province is visualized using ordinary kriging interpolation. The results reveal that the high-risk and extremely high-risk zones collectively cover 38,925 km², representing 27.31% of the provincial territory. The proposed BSEM framework offers an innovative methodology for assessing grid risks during rainstorm disasters, providing practical insights to enhance power sector resilience via improved early warning systems, resource optimization, and infrastructure planning.