International Journal of Scientific Research and Engineering Development

The efficiency and reliability of the modern power grid can generally be attributed to this importance to the infrastructure and economy of countries. Conventional maintenance strategies of either reactive or scheduled type cannot stop the unexpected failure of equipment easily, hence they result in expensive downtime and service outages. In this paper, the authors discuss the importance of using AIdriven predictive maintenance models to implement power grid operations to solve these issues. The proposed framework implements the latest machine learning models (Random Forest, LSTM, CNNs) to Process real-time and historical data to identify anomalies and predict possible failures prior to their happening. As shown in the case study, the predictive system leads to better accuracy of fault detection, reduced maintenance costs, and grid resiliency in general. Also, a deployment architecture is suggested to implement the solution in the edge and cloud context. Not only does it introduce a scaleable AI architecture, but the study also provides practical implications to grid operators to streamline maintenance plans and improve operational continuity.