基于SCADA系统的风电变桨故障预测方法研究  

Variable pitch fault prediction of wind power system based on SCADA system

作　　者：肖成[1,2] 刘作军[1] 张磊[1] 

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

出　　处：《可再生能源》2017年第2期278-284,共7页Renewable Energy Resources

基　　金：国家自然科学基金项目(61174009);河北省教育厅青年基金项目(QN2016104);廊坊市科技局项目(2015011033);北华航天工业学院青年基金项目(KY-2015-02)

摘　　要：风电机组变桨系统故障是目前造成机组停机的第一原因。文章对未来30 min的风电机组变桨故障进行预测,通过分析变桨系统潜在故障,制定维护保养计划;针对兆瓦级风电机组,分析SCADA系统的数据,提取变桨距系统故障特征;从风速、风向、桨距角和电机转速的输入、输出关系出发,应用多元线性回归分析和BP神经网络分别对变桨系统进行模型训练,对比两种算法的预测能力。通过分析故障预测模型性能指标、误差指标和输出数据图形可知,BP神经网络在风电变桨系统中的故障预测效果优于多元线性回归预测。Variable Pitch control system is one of the core control technology of wind turbine generator. Its failure has become the first reason that causes the unit to shut down. Therefore ,this paper predicts the variable pitch fault of wind power system in the future 30 minutes. It has a good reference for the analysis of the potential failure of the pitch system and the development of the maintenance plan. The data of SCADA system for MW class wind turbine is analyzed,and the fault characteristics of pitch control system i~ extracted. Multiple linear regression analysis and BP neural network model are used to train on the pitch system, and wind speed, wind direction, pitch angle and the relationship between input and output of motor speed are considered. The predictive power of the two algorithms are compared. And it compared with the fault prediction model performance index, error index and output data graph. Through the comparisons shown that the BP neural network prediction model is better than multiple regression orediction model.

关 键 词：变桨系统 故障预测 多元线性回归分析 BP神经网络 SCADA系统 

分 类 号：TK83[动力工程及工程热物理—流体机械及工程]