采用信息熵与组合模型的风电机组异常检测方法.pdfVIP

第39 卷 第3 期 电 网 技 术 Vol. 39 No. 3 2015 年3 月 Power System Technology Mar. 2015 文章编号：1000-3673 （2015 ）03-0737-07 中图分类号：TM 721 文献标志码：A 学科代码：470·4051 采用信息熵和组合模型的风电机组异常检测方法 颜永龙，李剑，李辉，孙鹏，张晓萌 （输配电装备及系统安全与新技术国家重点实验室(重庆大学) ，重庆市 沙坪坝区 400030 ） A Wind Turbine Anomaly Detection Method Based on Information Entropy and Combination Model YAN Yonglong, LI Jian, LI Hui, SUN Peng, ZHANG Xiaomeng (State Key Laboratory of Power Transmission Equipment System Security and New Technology(Chongqing University), Shapingba District, Chongqing 400030, China) ABSTRACT: At present, it is the most rapid and effective way 的风速范围；其次，建立了分别基于反向传播神经网络和最 to enhance operating reliability of wind turbines (WT) by 小二乘支持向量机的单项模型，分析了各单项模型的权重分 making full use of the information from supervisory control 布，并建立了较优的组合预测模型；最后，引入信息熵的概 and data acquisition (SCADA). Firstly, according to the 念对残差数据进行了处理，提出了风电机组状态参数的异常 SCADA data of wind farm the state parameters of WT and the 检测方法。验证结果表明，所提方法能准确检测出不同风电 range of the monitored wind speed are determined; secondly, 机组的发电机轴承温度异常，同时为风电机组的异常识别、 the individual model based on back-propagation neural 故障预警奠定了基础，可为风电场工作人员制定维修策略提 network (BPNN) and that based on least squares support vector 供参考信息，具有重要的工程价值。 machine (LSSVM) are established respectively and the weight 关键词：风电机组；数据采集与监控；状态参数建模；组合 distribution in each individual model is analyzed, and then an 模型；信息熵；异常检测 optimized combination forecast model is built; finally, the