摘要
针对风电机组齿轮箱故障预警问题,提出一种基于MIC-(PSO-BP)-MWA-KDE的齿轮箱油温预警方法。首先,使用最大互信息系数(Maximal Information Coefficient,MIC)求出与齿轮箱油温相关性高的参数作为模型的输入,采用PSO-BP神经网络构建齿轮箱油温预测模型。然后,通过计算齿轮箱油温实际值与预测值的残差绝对值,结合移动加权平均法(Moving Weighted Average,MWA)、核密度估计(Kernel Density Estimation,KDE)建立齿轮箱状态监测模型。通过实际案例分析可知,本文提出的预警方法可提前齿轮箱油温异常预警时间,预警时间提前约11小时。
Aiming at the problem of gearbox fault warning of the wind turbine,a gearbox oil temperature fault warning method based on MIC-(PSO-BP)-MWA-KDE is proposed.Firstly,the parameters with a high correlation with gearbox oil temperature are obtained by using the maximum information coefficient(MIC)as the input of the model.And the PSOBP neural network is used to construct the gearbox oil temperature forecasting model.Then,the gearbox condition monitoring model is established by calculating the absolute residual value of the actual value and predicted value of gearbox oil temperature,combined with moving weighted average(MWA) and kernel density estimation(KDE).Through the real case analysis,it can be seen that the fault warning method presented in this paper can advance the early warning time of abnormal gearbox oil temperature by about 11 hours.
作者
彭丽君
刘绪军
PENG Li-jun;LIU Xu-jun(School of Big data and Computer,Jiangxi University of Engineering,Xinyu 338000,China)
出处
《电脑与信息技术》
2022年第3期49-52,共4页
Computer and Information Technology
关键词
风机齿轮箱
状态监测
故障预警
神经网络
wind turbine gearbox
condition monitoring
fault warning
neural network
