Predictive analytics for 11kV underground cables: leveraging advanced analytics and machine learning to reduce breakdowns and operational costs

Tenaga Nasional Berhad (TNB), Malaysia's leading utility provider, is leveraging advanced analytics and machine learning to address the operational challenges associated with its extensive network of 11kV underground cables. Recognizing the critical need for predictive maintenance (PdM), TNB has developed a model to enhance cable reliability and optimize resource allocation. This model analyzes historical data, including insulation resistance (IR) test history, maintenance records, breakdown histories, cable length, age, and geospatial information. The model, trained on data from 2018 to 2022, utilizes advanced feature engineering methods to manage complex, high-dimensional data and addresses data imbalances to enhance predictive accuracy. A pilot program in Selangor region demonstrated an 80% accuracy rate in predicting cable failures within a two-year timeframe. This success has resulted in significant cost savings and improved network reliability. The model is currently being scaled nationwide, involving four key phases: Ideate, Explore, Incubate, and Nationwide Scale. Future enhancements include integrating real-time data, such as Partial Discharge (PD) mapping and load data, as well as incorporating additional geospatial data for deeper insights. TNB's predictive model represents a significant step toward a data-driven approach to asset management, ultimately aiming to improve grid reliability and customer satisfaction.