基于神经网络和小生境遗传算法的风电场短期功率预测研究-电力系统及其自动化专业论文.docxVIP

II II 和预选择算法相结合的算法可以得到较好结果，得出最终适应度值。在选 取的规律性最强的部分为风速段 4-16（m/s）的部分，误差带呈现带状分布， 误差分布在-30%至 30%段最为集中。其中也有一大部分数据误差异常，表明 风电预测呈现的间歇性，随机性以及不稳定性非常强。 关键词：风电 功率预测 BPNN RBF 小生境遗传算法 III III RESEARCH OF SHORT-TERM WIND POWER PREDICTION BASED ON NEURAL NETWORK AND NICHE GENETIC ALGORITHM ABSTRACT Wind power is a very HYPERLINK dict://key.0895DFE8DB67F9409DB285590D870EDD/universal universal power generation technology in recent years. China wind power technology has come to large-scale development stage. Because of its intermittence, in HYPERLINK dict://key.0895DFE8DB67F9409DB285590D870EDD/stability stability and hard- HYPERLINK dict://key.0895DFE8DB67F9409DB285590D870EDD/predictability predictability character, especially when it parallels in the whole grid, which can bring great influence to the stability and safety to the whole power grid. This is the bottleneck of wind-power development at present, also it is a challenging problem. In order to solve the problem of wind power, it is necessary to predict wind power. In this paper, wind power outputting characteristics are summarized and conducted related prediction in wind farm , the main texts as follows: Analysis the background and current situation at home and abroad, mastering the wind speed and wind turbine models. Through the forecasted wind speed on Back-Propagation Neural Network (BPNN) and Radial-Basis Function (RBF) prediction methods, combining with wind speed and power, the paper conducted wind -power prediction. Another is directly power prediction based on the speed and power data. Wind power training mean square error obtained MSE=0.3375 by BPNN, wind power predicting mean square error obtained MSE=0.4239 by BPNN. Wind speed training mean square error obtained MSE=0.0969 by BPNN, wind speed predicting mean square error obtained MSE=0.0169 by BPNN. Wind power predicting mean square error obtained PAGE PAGE VI MSE=0.0595 by RBF. Based on Niche Genetic Algorithm, the writer conduct wind power forecasting analysi