Improved Mid-long Term Load Forecasting Model Based On Intelligent Algorithm

China's Power demand nowadays is faced with great uncertainties due to various complex factors. The traditional load forecasting model based on fixed structure and parameters can no longer meet new load forecast requirements. So load forecast studies are gradually focused onto how to consider various uncertainties and achieve intelligent forecasting.This thesis is intended to take use of intelligent algorithm's advantages over handling uncertainties and adaptability to improve the existed load forecasting models, i.e. traditional system dynamics forecasting model and variable-weight combination forecasting model. The main content of this thesis is as follows:1,First, it gives a brief introduction to the existed load forecasting models, especially the traditional system dynamics forecasting model and variable-weight combination forecasting model. Then, it discusses the intelligent algorithm, i.e. Genetic Algorithm (GA) and Group Method of Data Handling (GMDH)'s advantages over improving the traditional load forecasting models. Plus, the uncertainty theory is also discussed.2,Due to GA's great efficiency on structural optimization and parameter estimation, the improved SD model based on GA was proposed to avoid traditional SD model's deficiency on adaptability, structure, and parameter sensitivity. It greatly expands the application range of SD load forecasting models.3,Intelligent combination forecasting model based on GMDH weighting method is proposed due to GMDH's efficiency on nonlinear system modeling. It is a great attempt of nonlinear variable-weighting method. Furthermore, variable-weight error correction mechanism is applied to deal with uncertainties to achieve flexible and intelligent forecasting, thus is good example for the study of intelligent mid-long term load forecasting modeling under uncertainties.4,Through programming with Vensim PLE 5.0 and MATLAB 7.0, the example of China's regional power grid verifies the feasibility of the proposed models. The load forecasting results and uncertainty analysis (structure adjustment and technical progress) of the 12th 5-year plan period indicates the rapid growth of future regional power demand and factor (technical progress)'s positive effects to China's energy conservation.In a word, the models proposed in this thesis bear greater flexibility and intelligence in consideration of various uncertainties. They are especially fit for nowadays'china, and can arouse great possibility for future's intelligent load forecasting studies under uncertainties.