用于风力发电仿真的多时间尺度风速建模方法.pdfVIP

第 37 卷 第 9 期 电 网 技 术 Vol. 37 No. 9 2013 年 9 月 Power System Technology Sep. 2013 文章编号：1000-3673 （2013 ）09-2559-07 中图分类号：TM 614 文献标志码：A 学科代码：470·40 用于风力发电仿真的多时间尺度风速建模方法 佘慎思，李征，蔡旭 （上海交通大学风力发电研究中心，上海市 闵行区 200240 ） A Multi-Time Scale Wind Speed Modeling Method for Simulation of Wind Power Generation SHE Shensi, LI Zheng, CAI Xu (Wind Power Research Center, Shanghai Jiao Tong University, Minhang District, Shanghai 200240, China) ABSTRACT: To simulate output characteristics of wind power 力系统动态过程和中长期经济调度问题具有非常重要的意 generator a multi time scale wind speed model is established, 义。采用分段线性化方法和基于夹角距离的相似度匹配进行 and it is very important for the study on dynamic process and 风速的同类趋势聚合，利用聚合数据构建径向基函数神经网 medium- and long-term economic dispatching of power grid 络拟合风速趋势，并在此基础上结合风速集合量化概率分布 containing wind farm to build effective steady-state and 和趋势变化 2 个约束，对概率分布函数生成的随机风速序列 transient models of wind farm to research the operating 按照趋势的最小偏度进行重构，建立符合真实变化规律的风 conditions of power grid integrated with wind farm by 速模型。算例基于一组实测数据建立了同趋势的随机风速模 simulation analysis. Using piece-wise linearization and 型，从变化曲线和功率谱密度曲线对真实风速序列和模拟的 similarity matching based on angle distance the wind speed 随机风速序列进行了分析对比。实验结果证明，与传统的建 with similar trend is aggregated and using the aggregated data a 模方法相比，建立的模型既能够反映风速的长期趋势特征， radial basis function neural network (RBFNN) is constructed to 又保留了风速变化的动态特性，显著提升了风速模型的多样 fit the trend of wind speed, and on this basis of combining with 性与真实性。 the two constraints, i.e., the quantized probability distribution 关键词：风速建模；时间尺度；径向基函数神经网络；概率 of wind speed aggregation and the trend variation, the random 分布；功率谱密度 wind s