Peak Load Forecasting Based on Long Short Term Memory

Electricity load forecasting is a very imperative issue not only in the power industry but also one of the factors to expand the economic efficiency of power and integral to the plan and execution of various vital. In this case, we use a machine learning approach, exclusively, Long Short Term Memory (LSTM) for predicting future the peak load based on historical data which recorded from Sub-Station in Lhokseumawe, Indonesia. LSTM is capable of forecasting complex univariate electric load time series with strong no-stationarity. The result shows the effectiveness of LSTM and outperform traditional forecasting methods in the challenging peak load record problem.