Electrical Energy Forecasting of Industrial Feeder from Rural Area Using Different Regression Techniques

We are aware about the role of machine learning in engineering domain increasing day by day to ease the life of human being. Now a day's population, globalization and industrialization are increasing which lead to increasing electrical consumption and due to this challenges for existing energy infrastructure increased. Entire world looking for the cheap, smooth, easily available and reliable energy resources to fulfill this challenges in efficient way at local level so energy generation and consumption will match to each other. To minimize carbon and other contaminant emissions, there is combination of renewable and clean distributed generation energy is increasing continuously. But integrating these energy into the grid may increase in electrical demand consumption enhance instability. With the help of Electrical load forecasting; we can reduce the grid instability and enhances the strength of power quality. Electrical energy forecasting by referring a historical data modelling plays a major role in base, peak electrical energy consumption, generation balancing; but forecasting an accurate electrical energy data is very tough work due to variation and nonlinearity in load nature. Machine Learning has different forecasting methods based on the previous data set. In this study a dataset is gathered from Industrial Feeder in Nasik rural district, Maharashtra state, India. The aim of this real time case study is to design a model to forecast the electrical energy consumption by using different regression techniques like LR, SVM, LSTM, KNN, Bi-layered NN etc. A best fitted model using KNN is identified based on the evaluation parameters values.