Long-Term Electrical Load Forecasting of Wolaita Sodo Town, Ethiopia Using Hybrid Model Approaches

The problem of imbalance power in Wolaita Sodo town due to the time lag between awareness of future load and satisfying that load is addressed using long term electrical load forecasting. Hybrid model of Multi-variable linear regression with artificial neural network and Hybrid model of Multi-variable linear regression with adaptive neuro-fuzzy inference system are used in this paper. To demonstrate the effectiveness of the proposed approaches, past electrical load, population growth and gross domestic product of nine-year data is taken to forecast the electrical load of the town for the six years ahead. The different models or techniques are compared based on some error performance criteria. The forecasted result has shown that the electrical load consumption of the town would be likely to increase from 13.568 MW for the year of 2017 to 22 MW in 2023. Since the maximum capacity of the present substation supplying power to the town is 16 MW; extra 37.5% megawatt is required after six years. From the two major factors, population has more share than gross domestic product for the increase of electrical power in the town in each year.