The main objective of this study is to investigate and predict short-term electrical load usage and consumption in the metropolitan hub Karachi, a city of Pakistan. This study uses 10 sets of monthly data of 100 home users distributed over a period of 10 years as an input to forecast load consumption for the forthcoming month of October, next year. In this work a comparative study is presented, prediction models have been developed using regression analysis and artificial neural network (ANN). The models in consideration utilize different attributes as inputs and forecast future load consumption as an outcome. Data used to train the models have been taken from the meteorological department, government of Pakistan and from K-Electric corporation (KE), Karachi. It is of a power plant serving a specific area in the city of Karachi. Moreover, the aforementioned datasets are proprieties of KE and meteorological department. Further, parameters used to train the model, includes, weather, temperature, wind speed, dew factor, relative humidity and air pressure. Forecasting was done using regression analysis (RA) and multi-layer feed-forward neural network (MFNN) via back-propagation algorithm. Results and convergence of both the aforesaid models were compared. Daily peak loads were analysed for a period of 31 days for the month of October for a duration of 10 years from 2008 to 2017. To measure performance and level of accuracy of the predicting models, simulation results were compared with actual data for the month of October 2018. Furthermore, MFNN has shown to outperform the regression model in terms of prediction accuracy. Moreover, MFNN provided best results with an accuracy level of up to 96%. In addition, among all the input attributes, temperature was found to be the most relevant to model's good performance in terms of prediction accuracy.
[1]
Hany M. Hasanien,et al.
Short term load forecasting using ANN technique
,
2017,
2017 Nineteenth International Middle East Power Systems Conference (MEPCON).
[3]
Abbas Khosravi,et al.
Short-Term Load and Wind Power Forecasting Using Neural Network-Based Prediction Intervals
,
2014,
IEEE Transactions on Neural Networks and Learning Systems.
[4]
S. A. Soliman,et al.
Long-term/mid-term electric load forecasting based on short-term correlation and annual growth
,
2005
.
[5]
Norman Chung-Fai Tse,et al.
Short-term load forecasting coupled with weather profile generation methodology
,
2018
.
[6]
Ajay Shekhar Pandey,et al.
Short-term load forecasting methods: A review
,
2016,
2016 International Conference on Emerging Trends in Electrical Electronics & Sustainable Energy Systems (ICETEESES).
[7]
Ying Chen,et al.
Short-Term Load Forecasting: Similar Day-Based Wavelet Neural Networks
,
2010,
IEEE Transactions on Power Systems.
[8]
Hossein Sangrody,et al.
Long term forecasting using machine learning methods
,
2018,
2018 IEEE Power and Energy Conference at Illinois (PECI).
[9]
Rob J Hyndman,et al.
Short-Term Load Forecasting Based on a Semi-Parametric Additive Model
,
2012,
IEEE Transactions on Power Systems.