LOAD FORECASTING USING AI TECHNIQUES

Abstract

Electricity is the greatest gift of science to mankind. World have reached a point of civilization when electricity is used for all purposes. In India, the demand for electricity is increasing day by day. As on 30th September 2021, the total installed capacity is 3,88,848 MW of India. The reason for increase in consumption of electricity is due to urbanization, increasing population. It can be concluded that in the upcoming time this demand will keep on increasing. Electricity is generated on the basis of demand. It is, consequently, basic for the electric force utilities that the heap on their frameworks ought to be assessed ahead of time. This assessment of burden ahead of time is ordinarily known as load forecasting. As different constraints of regular methodologies, the accentuation has gradually moved to the use of Artificial Intelligence based methodologies. Fuzzy logic speculation is one of transcendent advancement in artificial intelligence. Its application in load estimating depends on periodical comparability of electric burden, where the information factors, yield factors and rules are the central issue. Load forecasts are additionally used to set up obtainment arrangements for development capital energy estimates, which are expected to decide future fuel necessities. Hence, a decent conjecture, mirroring the present and future patterns, is the way in to all arranging. The term forecast alludes to projected burden prerequisites decided utilizing an efficient cycle of characterizing future needs in adequate quantitative detail to allow significant framework extension choices to be made. Dismally, the customer load is basically disorderly albeit minor varieties can be influenced by frequency control and all the more radically by load shedding. The variety in load shows certain every day and yearly example. This dissertation work “Load Forecasting using AI Techniques”, focus is on short term load forecasting which is substantial for controlling and operation in real time of the power system. The proposition is the use of artificial intelligence techniques like fuzzy logic, artificial neural network and adaptive neuro-fuzzy inference system. All three models are studied for the set of data considered and the results are analysed. Also, the results obtained from above methods is compared to the desired output and the mean absolute percentage error is calculated.