摘要
采用风电机组状态监测技术可有效提高机组运行的安全可靠性。轴承是风电机组能量传递的重要部件,轴承的状态评估对机组安全运行具有重要意义。文章基于主成分分析方法,选取影响机组轴承温度的参数,提出了改进的线性回归径向基函数神经网络方法,建立了正常运行状态下轴承的温度预测模型;通过机组运行数据的分析比较,采用滑动窗口残差统计方法对机组运行状态进行实时监视评价发现,发电机出现异常时,轴承温度呈现上升趋势,残差值超过设定的置信区间,从而能实现对故障的有效预测。文章的研究结果可为风电机组的安全高效运行提供参考。
Condition monitoring of wind turbines is important for ensuring the safety and stabilization of turbines,the bearing is a key part of energy transfer,whose condition estimation is signficant for turbines. A better linear regression RBF algorithm was applied to constructing the normal behavior model of the bearing temperature on account of principal component analysis. On this basis,the temperature was monitored according to the data in Supervisory Control And Data Acquisition system,with the aid of sliding window statistics. The results shows that when the generator is abnormal,an upward trend of the temperature is shown,and the residual value would exceed the setted threshold interval,the fault is predicted. This study can provide a reference for the safe and efficient operation of wind turbines.
出处
《可再生能源》
CAS
北大核心
2018年第2期276-282,共7页
Renewable Energy Resources
基金
国家自然科学基金项目(51507053)
中央高校基本科研业务费项目(2017B42314)
关键词
风电机组
轴承温度
线性回归RBF神经网络
残差
故障预测
wind turbine
bearing temperature
linear regression RBF neural network
residual
fault prediction
