Abstract: With the continuous advancement of substation automation and operation technologies, the real-time requirements for multi-source information interaction have increased significantly. Traditional anti-misoperation systems suffer from rigid logic, delayed response, and insufficient risk perception, making them inadequate for the operational demands of modern power grids. To address these challenges, this paper proposes an intelligent anti-misoperation method for substations based on a cloud-edge collaborative architecture. By constructing a cooperative model integrating cloud-based strategy optimization with edge-side real-time decision-making, the proposed method enables dynamic learning of anti-misoperation logic and rapid interlock response. Furthermore, a substation anti-misoperation feature model, dynamic deviation evaluation function, and confidence fusion algorithm are developed to form a multidimensional data-driven judgment framework. Pilot verification conducted at an actual substation demonstrates that this method significantly outperforms traditional anti-misoperation logic in terms of operation recognition accuracy, response latency, and system stability, effectively enhancing the intelligence and safety level of substation protection.