Industrial electricity consumption forecasting

Abstract

Forecasting future industrial electricity consumption is critical to the Electric power production company and Industrial factory. Both energy planning and industrial production need forecasting data as a reference. Therefore, this thesis will introduce and demonstrate the forecasting of the future industrial electricity consumption data. This thesis use an Long Short Term Memory (LSTM) Time Series model to predict power consumption data. This model can transmit information on time series data and also can handle long-term-dependency issues. These features are ideal for predicting industrial electricity consumption. The predicted objects and training data come from a small paper mill in Chile. Although the data is processed for confidentiality reasons it can still be used for machine learning. Finally, this thesis will use the created LSTM model to predict the industrial electricity consumption of this factory in January of the next year.