To address industry challenges in the remote operation and maintenance (O&M) of power system secondary equipment, such as disordered alarm information in traditional protection and control systems, large errors in secondary circuit fault localization, high misjudgment rates in manual monitoring, and significant deficiencies across the entire “perception-cognition-decision” chain, an intelligent monitoring system for secondary equipment is developed. This system integrates precise alarm classification and grading, intelligent identification of equipment operation scenarios, and automated generation of alarm handling measures, along with the development of a corresponding remote O&M management platform. The intelligent monitoring system standardizes alarm information using natural language processing and pretrained models, while establishing a dynamic iterative optimization mechanism. It relies on a multidimensional, constraint-based scenario knowledge base and a rule-based reasoning engine to achieve accurate identification of equipment operation scenarios. By combining multi-strategy maintenanceinformation filtering with large language models and retrieval-augmented generation (RAG) techniques, the system generates effective alarm handling solutions. The platform supports functions such as remote monitoring, inspection, control, and maintenance, and is designed based on a secure zoning architecture. It has been deployed in a provincial power grid and has demonstrated stable performance, promoting the transition of secondary equipment O&M from “on-site dependence” to “remote intelligence”.