Short Term Load Forecasting Of Urban Loads Based On Artificial Neural Network

Abstract

This paper illustrates the growth and utilization of neural networks towards a thriving topic of modern electrical studies- ‘short-term load forecasting’. Load forecasting/estimation is of great value to Power system studies and its inspection. In this paper, the study of short-term load forecasting concerning the Roorkee region situated in the north of India (Uttarakhand state) was explored. Historical data in the form of load flow in hourly rate was provided to the Artificial Neural Network from the previous set of days. Hourly-rate forecast has been done for optimally estimating the load flow in the given region. The comparison of estimated load flow is done with the real time data to get the measure of its accuracy. Thus, acceptable load estimations have been done without the added complexity of weather data interference. The obtained results convey the sustainability /suitability of the methodology used in the paper for short-term load forecasting. The design includes selecting a resembling day load as the input load and use Artificial Neural Network for predicting the value of load at any particular time of the day. The performance of the neural network is tested through an extensive study of the hourly-based data from the Roorkee region. The estimated output shows the hourly load utilisation in the region. The data has been predicted for the fore coming week from that of the historical data