Short-Term Load Forecasting for Industrial Enterprises Based on Long Short-Term Memory Network

Abstract

In China, industrial enterprises develop rapidly, and industrial electricity consumption account for 70% of social electricity consumption. However, the charging mode of industrial electricity is very complicated which can be divided into two basic modes, one mode is to pay according to transformer capacity and other mode is to pay according to maximum demand. In both modes, the complex charging type involving seasonal factors and national economic growth. Once a company chooses the wrong mode to buy electricity, it has to change it three months later. In this case, it not only increases the cost of electricity but also wastes resources. Therefore, it is important to predict the future load data according to the energy consumption characteristics of enterprises. In the context of big data, the electricity data of Chinese enterprises will be developed and utilized gradually. In this paper, the long short-term memory (LSTM) model is built to carry out load forecasting of the international dairy companies which called Ausnutria. The simulation results show that LSTM network is feasible in the field of load prediction, and the method can ensure the accuracy.