Short Term Electric Load Forecasting of 132/33KV Maiduguri Transmission Substation using Adaptive Neuro-Fuzzy Inference System (ANFIS)

Abstract

This article provides a way of accurately predicting one-hourahead load of a utility company located in the North Eastern region of Nigeria based on Adaptive Neuro-Fuzzy Inference System (ANFIS). The inputs to the ANFIS are the next-hour temperature, next-hour humidity, day of the week, hour of the day, and the current-hour load. The output is the next-hour load of the entire system. All the data used span the period 2009 to 2012 (4 years). These parameters are non-linear, stochastic (random) and uncertain in nature. Adaptive Neurofuzzy based Inference System (ANFIS), an integrated system, comprising of fuzzy logic and Neural Network was used to model the next hour load, because it can address and solve problems related to non-linearity, randomness and uncertainty of data. 75% of the data was used for training and 25% for checking. From the analysis carried out on the ANFIS-based model; Mean absolute percentage error (MAPE) for a typical Monday, Wednesday and Friday was found to be 12.61%, 12.76% and 12.12%. The Mean absolute error (MAPE) on the entire test data was 24.76%. The analysis shows satisfactory level of accuracy with regards to the ANFIS-based model developed in forecasting the next hour load especially with a correlation (r) value of 84.64%. General TermsLoad Forecasting.