Research And System Implementation Of Mid-Long Term Load Forecasting Based On Combination Weight

Power grid planning is the basis and the prerequisite of power grid construction, the quality of planning significantly affect the security and stability of power grid operation. Mid-long term load forecasting is the foundation of power grid planning, it not only provides necessary and basic data for power grid planning, while its forecast accuracy has a direct impact on the reasonableness of investment and operation of the planning, which related to the safety,the economical and reliable operation of the power grid.There are many kinds of load forecasting methods, but each method has its scope and limitation, which may not perform well in any situation. Combination forecasting is an effective strategy for forecasting, it can integrate the results of different forecasting methods, and maximize the use of information from various forecasting samples, which will improve the prediction result. It is important and difficult for combination forecasting models that how to choose the appropriate models from a number of forecasting models, and how to give appropriate weight to each selected model.In point of the choosing of single model for combination forecasting method, this paper presents a multi-index comprehensive evaluation method based on gray correlation analysis and TOPSIS method. First, select a variety of indicators which include error and associated indicators, and use them to construct ideal solution and negative ideal solution. Then calculate the comprehensive relative closeness between each forecasting method and the ideal model, and sorting them according to the comprehensive relative closeness of each forecasting method. Finally, screening the models based on the setting threshold. Examples demonstrate the feasibility and effectiveness of this method.In point of the weight determination issue of combination forecasting, this paper presents a mid-long term load combination forecasting model which is based on the improved comprehensive validity indicator. First improve the Grey Relational Grade and the prediction validity of each forecasting method, using objective and subjective weighting methods to impart appropriate weights, and then calculating the improved comprehensive validity indicator. Finally, giving appropriate weights to each forecasting method according to improved comprehensive validity indicator, and get the final forecasting result.On this basis, combining the actual demand of power grid planning, an electric power demand forecasting and load characteristic analysis platform was developed, which provided technical support for the power grid planning, meeting the application demand of power grid planning in power demand forecasting effectively.