Title: Comparative Analysis Of Ann And Anfis Load Forecasting Techniques At Okuru Community, Rivers State
Authors: Blessing Dike, Bodise Lebrun Bolou-sobai
Volume: 10
Issue: 8
Pages: 68-74
Publication Date: 2026/08/28
Abstract:
Precise electricity load forecasting is essential for power system planning in developing areas with growing electricity demand due to population growth and urbanization. This research compares the Artificial Neural Network (ANN) and Adaptive Neuro-Fuzzy Inference System (ANFIS) methods to predict the consumption in Nigeria's Okuru 33kV Substation in Nigeria using historical data for the period 2010 to 2022. The ANFIS model used grid partitioning, triangular membership functions, and a hybrid optimization algorithm. The ANN was a feed-forward type back-propagation network, and 10 hidden neurons were used, with the Levenberg-Marquardt algorithm. Evaluation of performance was done based on Mean Squared Error (MSE) and regression coefficient (R). ANFIS had a lower training MSE (0.208) than ANN (0.320), and ANN had a very good training R-value (0.916), validation R-value (0.924), and testing R-value (0.952), indicating good predictive ability. The rising trend in consumption was captured by both models, increasing from 2.8 MW in 2010 to 4.5 MW in 2022. However, ANFIS showed high accuracy because of its better nonlinear modeling. Its forecast predicts growth from 5.396 MW in 2023 to 8.524 MW by 2032. The results of the study have shown that ANFIS is a reliable model for predicting demand, which can help in distribution planning, infrastructure development, integration of renewables, and sustainable energy management.