FORECASTING OF LONG-TERM ELECTRICAL LOAD DEMAND USING RECURRENT NEURAL NETWORK

Electrical load demand forecasting is an essential process in power system planning and operation, enabling utilities to optimize resource allocation, ensuring grid stability and facilitating demand-side management (DSM). This study employs a Recurrent Neural Network (RNN) to estimate the electricity demand in Ado-Ekiti Metropolis, Nigeria, for the period 2021 to 2030. The model was developed and simulated using MATLAB 9.0.0 (R2016b) curve fitting tools to establish the effectiveness of RNN in load forecasting. The RNN model was implemented as a nonlinear system utilizing a backpropagation learning algorithm to capture complex temporal dependencies in electricity consumption. The results indicate that the average annual predicted energy consumption over ten-year period is 1127 MW. Findings reveal a consistent increase in energy demand at the 11kV injection substation, particularly for Adebayo and Agric feeders, highlighting the necessity for proactive energy planning. The study emphasizes that accurate load forecasting helps electricity providers optimize power distribution, reduces operational losses and supports pricing mechanisms for potential investors. Integrating demand forecasting with generation planning is essential for maintaining a reliable power supply in Nigeria’s evolving electricity market. By leveraging deep learning techniques like RNN, utilities can enhance their forecasting capabilities, improve grid efficiency and ensure sustainable energy management.