Residential Electrical Long-Term Load Forecast Using Artificial Neural Network; Woji Estate 11/0.415 KV Feeder Port Harcourt, Nigeria
Publication Date: 30/01/2024
Author(s): Ogbonna Bartholomew Odinaka , Agboola Olasunkanmi Johnson.
Volume/Issue: Volume 6 , Issue 1 (2024)
Abstract:
The essentiality of electric load forecast for the effective design and management of electric power systems has been achieved in this study. PHEDC may plan for infrastructure construction, resource allocation, and energy management by using accurate long-term load forecasts of this study. In the context of the Woji Estate 11/0.415 kV Feeder in Port Harcourt, Nigeria, we have discussed the use of artificial neural networks (ANNs) for a long-term of ten (10) years load forecasting in this paper starting from January 2020- December 2029. However, curve fitting feed-forward artificial neural network has been employed for the simulation on MATLAB 2020 environment, with six(6) input datasets obtained from Transmission Company of Nigeria(TCN), Oginigba and Port Harcourt electricity distribution Company, and average temperature dataset from NIMET-Abuja all in Nigeria from January, 2015-December, 2019. The regression plot of epoch 11 with training; R=1 and validation of 0.9999 have been achieved which indicates how efficient was the training of the dataset. Levenberg-Marquardt (LM) algorithm is used as an optimization technique in this study. In addition, training, test, validation, and error analysis have been used to examine the effectiveness of the LM algorithm, it has been found optimal with ANN. The general observation shows that ANN provides effective results on long-term electrical load forecasting of the Woji Estate Feeder with a total forecasted value of 29734.4 MWHR and an average value of 24778.67 MWHR at the end of the tenth year.
Keywords:
The essentiality of electric load forecast for the effective design and management of electric power systems has been achieved in this study. PHEDC may plan for infrastructure construction, resource allocation, and energy management by using accurate long-term load forecasts of this study. In the context of the Woji Estate 11/0.415 kV Feeder in Port Harcourt, Nigeria, we have discussed the use of artificial neural networks (ANNs) for a long-term of ten (10) years load forecasting in this paper starting from January 2020- December 2029. However, curve fitting feed-forward artificial neural network has been employed for the simulation on MATLAB 2020 environment, with six (6) input datasets obtained from Transmission Company of Nigeria (TCN), Oginigba and Port Harcourt Electricity Distribution Company, and average temperature dataset from NIMET-Abuja all in Nigeria from January, 2015-December, 2019. The regression plot of epoch 11 with training; R=1 and validation of 0.9999 have been achieved which indicates how efficient the training of the dataset was. The Levenberg-Marquardt (LM) algorithm is used as an optimization technique in this study. In addition, training, test, validation, and error analysis have been used to examine the effectiveness of the LM algorithm; it has been found optimal with ANN. The general observation shows that ANN provides effective results on long-term electrical load forecasting of the Woji Estate Feeder with a total forecasted value of 29734.4 MWHR and an average value of 24778.67 MWHR at the end of the tenth year.
