Pattern-Based Forecasting Monthly Electricity Demand Using Multilayer Perceptron

Abstract

Medium-term electric energy demand forecasting is coming a key tool for energy management, power system operation and maintenance scheduling. This paper offers a solution to forecasting monthly electricity demand based on multilayer perceptron model which approximates a relationship between historical and future demand patterns. Energy demand time series exhibit non-stationarity, long-run trend, cycles of seasonal fluctuations and random noise. To simplify the forecasting problem the monthly demand time series is represented by patterns of yearly periods, which filter out a trend and unify data. An output variable is encoded using coding variables describing the process. The coding variables are determined on historical data or predicted using ARIMA and exponential smoothing. As an illustration, the proposed neural network model is applied to monthly energy demand forecasting for four European countries. The results confirm high accuracy of the model and its competitiveness compared to other models such as ARIMA, exponential smoothing, kernel regression and neuro-fuzzy system.