Medium and Long-Term Coal Demand for Electricity Sector Forecasting Based on Improved Seasonal Adjustment Model

Abstract

The electrical coal consumption in our country presents non-stationary characteristics of seasonal periodicity and circular trend while seasonal adjustment can decompose this sequence into trend cycle element, season element and irregular component with practical economic meaning. Before seasonal adjustment, we need to eliminate the impact of outlier, workday and leap year in the sequence in original electrical coal consumption and then we can conduct decomposition on the trend cycle sequence after seasonal adjustment applying H-P filtering method. After that, we can select appropriate model to conduct electrical coal demand forecasting based on different characteristics like long-term trend, periodic cycle, seasonal factor and irregular component after decomposition. Through the empirical test of electrical coal consumption in our country for 192 months, the results indicate that the precision has been improved significantly in long-term electrical coal demand forecasting by using the improved seasonal adjustment model and method.