Long and Medium Term Load Forecasting Based on Grey Model Optimized by Cosine Function and Time Response Function

The article uses the cosine function and Euclidean distance function to optimize the traditional GM (1, 1) model. The cosine function processing the raw data columns, which could enhance the smoothness of the original series, and strengthen the general trend of the original series as well, effectively eliminate the shock interference of disturbed system. The principle of minimum Euclidean distance used to optimize the response function to determine the constants C, which could make better use of the original information. The improved model is applied to the tertiary industry in Beijing load forecasting, the results show that the improved model has higher prediction accuracy, and the improved model used to predict the next three years’ electricity consumption of tertiary industry in Beijing.