APPLICATION OF DEEP LEARNING ALGORITHMS FOR PREDICTING POWER SYSTEM FAILURES IN REAL-TIME DATA-BASED SMART GRIDS

Abstract

This study aims to analyze the application of deep learning algorithms in predicting electricity system failures on smart grids based on real-time data. The research method used is a qualitative method with an exploratory case study design, which was chosen to gain an in-depth understanding of operational processes, data management, and the application of prediction systems in the smart grid environment. The research was carried out on a smart grid-based electricity distribution management unit in the urban area of Greater Jakarta. The research informants consisted of five purposively selected people, consisting of distribution managers, power systems engineers, information technology specialists, data analysts, and operational decision makers, with consideration of their direct involvement in the management and utilization of real-time data. The results show that deep learning algorithms are able to improve the accuracy and timeliness of failure prediction compared to conventional approaches that are reactive. The developed model successfully recognizes latent disturbance patterns and supports condition-based maintenance decision-making. This study recommends the development of more integrative prediction models, testing on a wider network scale, and strengthening decision support systems to improve the reliability of smart grids.