基于小波BP神经网络的风电机组变桨系统故障预测  

Fault prediction of variable pitch system of wind turbine based on wavelet BP neural network

作　　者：肖成[1,2] 焦智[2] 孙介涛 张磊[1] 宋玉彬[2] 石莹[2] 

机构地区：[1]河北工业大学控制科学与工程学院,天津300130 [2]北华航天工业学院电子与控制工程学院,河北廊坊065000 [3]厦门大学自动化系,福建厦门361005

出　　处：《可再生能源》2017年第6期893-899,共7页Renewable Energy Resources

基　　金：河北省教育厅青年基金项目(QN2016104);河北省科技厅指令性项目(16210310D)

摘　　要：变桨故障是风电机组重要的停机故障之一,对变桨系统进行故障预测并提高预测精度,是风电开发的关键技术,不但保证电网安全运行而且减少运维成本。分析处理SCADA系统数据,提取相关联参数,即输出功率、风速、桨距角和转子转速。采用BP神经网络对系统进行模型训练,考虑到风电机组参数具有波动性、不确定性等,同时采用小波BP神经网络进行模型训练。建立变桨故障预测模型,预测未来15 d的变桨系统运行情况,用于制定合理的运维方案。通过MATLAB系统仿真研究,对比分析了预测模型性能指标、误差指标和输出数据图形,小波BP神经网络训练预测模型诊断精度比BP神经网络提高了17%,可信率提高了18%,诊断能力提高了15.4%,诊断误报率降低了17%。Variable pitch fault is one of the most important faults in wind power system. It is the key technology for the development of variable pitch control system to predict the fault and improve the prediction accuracy. It not only guarantees the safe operation of the power grid,but also reduces the cost of operation and maintenance. Use the data of SCADA can extract the associated parameters. It includes the output power,wind speed,pitch angle and rotor speed. BP neural network is used for the model system. Considering the fluctuation of wind power system parameters,uncertainty etc.,at the same time,the wavelet BP neural network is used to train the model. It can be predicted that the variable pitch control system can be operated in the future 15 days through the establishment of pitch fault prediction model. It is used to develop a reasonable operation and maintenance program. Through the MATLAB system simulation study,the paper analyzes the performance of the forecast model,the error index and the output data. Wavelet BP neural network training prediction model diagnostic accuracy than BP neural network increased by 17%,the reliability rate increased by 18%,the diagnostic ability increased by 15.4%,and the diagnostic false alarm rate was reduced by 17%.

关 键 词：风电机组 变桨系统 故障预测 BP神经网络 小波BP神经网络 

分 类 号：TK83[动力工程及工程热物理—流体机械及工程]