The Application Of Semi-parametric Regression Model In Long-middle Term Load Forecasting

Load forecasting is the basis for power system planning, and is also the cardinal foundation for ensuring power grid operation economic and safely and achieving the scientific management and scheduling of the grid. The improvement of technical level of power forecasting is beneficial for power plan utilization management and reasonable arrangements for operation mode and formulating a reasonable power source construction planning and the improving the economic and social benefits. Power load forecasting is the experts'major concern. Improving the accuracy of power load forecasting and expansion the scope of the load forecasting application are of great significance.Semi-parametric regression model including with regression model and non-parametric regression model are not the simple sum of the two models. It has the advantages of the parameters and non-parametric regression model focusing the information of the main component (parameter part), without neglecting the role of interference terms (non-parametric part).Because many factors affect the load forecast of its impacts are not the same, loss of data, data exception data, and the collinearity between factors has a greater influence on the load forecast, the raw data is processed and in accordance with the correlation coefficient and The total sum of squares of partial regression to select the independent variables. Using the two-stage least squares estimation method to evaluate the part of parametric and semi-parameter regression model and then residual test and the final analysis of the load forecasting. Examples show that: semi-parametric regression model fitting results is better applied to load forecasting. The high precision prediction expands the scope of application.