Research On Wind Speed Forecasting Based On Error Correction And Fuzzy Evaluation

In recent years, the increasing seriousness of global energy crisis, ongoing deterioration of ecological environment, the vision of people is attracted to adjust energy structure and look for clean, green, unlimited use of alternative energy. As a form of conversion of solar energy, wind is regarded as the most potential development and most widely use of green energy, with the advantages of safety, non-polluting, renewable, environmental protection and ecological benefits friendly. Wind power is the most common approach to utilizing wind, while the capacity of wind power integration exceeding a proportion will severely affect the power grid on account of the stochastic volatility. Therefore, wind speed forecasting in advance can relieve the negative influence of wind power integration by predicting the situation of variation of speed and output of fans.This paper analyzes the characteristics of distribution and variation of wind speed and the relationships of wind speed and its influence factors, and the method of error correction based on Markov Chain is put forward in the error correction perspective. Taking into account of the drawbacks for single evaluation index evaluating the performance of models, a comprehensive evaluation method based on fuzzy pattern recognition is proposed, providing a reference for wind speed forecasting decision-maker optimizing models.The main contributions of this paper is as follows:(1) The background and significance of wind speed forecasting are expounded, the current development status of wind power generation technology both in China and abroad, the methods of wind speed forecasting and approaches of evaluating models are summarized, the major problems of wind speed forecasting are analyzed.(2) The characteristics of wind speed distribution, season variation, monthly variation and daily variation are analyzed, the variation relation of wind speed with temperature, humidity, air pressured, history wind speed are researched, the error source of wind speed forecasting is summarized.(3) The method of error correction based on Markov Chain is put forward from the perspective of error correction to improve the wind speed forecasting accuracy. Firstly, fuzzy C-means clustering algorithm is employed to solve the problem of the state division of Markov Chain. Secondly, the steps of error correction are given in detail. Finally, the proposed method is utilized to correct the models of Generalized Regression Neutral Network based on Cross Validation, Elman neutral network based on genetic algorithm, improved T-S fuzzy neutral network. The comparison of results of three models for wind speed forecasting with different time scale before and after correction show that the proposed method can improve the accuracy and performance of wind speed forecasting.(4) A comprehensive evaluation method based on fuzzy pattern recognition is presented, aiming at solving the unreasonable and partial problems of single evaluation index evaluating the performance of models. On the basic of wind speed forecasting performance evaluation index system established by our team, the subjective weigthts of indexs are determined by dual contrast method, and the maximum deviation method is used to determine the objective weights. On the foundation of integrating subjective and objective weights, fuzzy pattern recognition method is utilized to evaluate the models of wind speed forecasting on the platform of MATLAB. The simulation results show that the rationality and effective feasibility of proposed method, providing a feasibility method for evaluation.