Publication

Abstract :

Mini-grids play a crucial role in powering off-grid and remote locations with electricity. However, they are plagued by issues like variability of load, irregularities of generation, and abrupt system breakdowns resulting in blackouts. The present paper proposes a novel artificial neural network (ANN) based method to predict blackout occurrences in mini-grids by utilizing real-time and past grid data. Our system continually observes grid performance, learning base patterns of instability and upcoming failure. With L2 regularization and dropout techniques implemented, the ANN is tuned for generalization robustness despite natural fluctuations in the conditions of the grid. In contrast to conventional prediction techniques based on pure static analysis, our approach adjusts itself dynamically in response to changing grid patterns by constant learning, facilitating remedial action in a timely manner and minimizing system downtime. The suggested system not only improves energy distribution but also ensures the sustainability and resilience of decentralized energy networks. This paper presents a major contribution to the field through the provision of a smart, adaptive solution for enhancing the resilience of mini-grids against interruptions, thus promoting sustainable development in remote communities,this model serves the purpose with 97.5 % accuracy.