摘要
为评估风电机组高速轴轴承的健康状况并预测其后续状态,构建能反映轴承退化过程的健康指数(HI)曲线作为健康状态预测的基础,提出一种核主元分析(KPCA)与双向长短期记忆(Bi-LSTM)网络相结合的方法。首先利用单调性及核主元分析构建高速轴轴承的HI曲线,其次在已构建的HI曲线基础上利用Bi-LSTM网络对高速轴轴承的健康状态进行预测。通过高速轴轴承全寿命周期试验,验证了该方法构建的HI曲线单调性好,能对高速轴轴承进行较为准确的健康评估和状态预测。
In order to evaluate the health condition of the wind turbine high-speed shaft bearings and predict its subsequent state,a health index(HI)curve reflecting the degradation process of the bearing was constructed as the basis of health condition prediction,A method based on Kernel Principal Component Analysis(KPCA)and Bidirectional Long Shot Term Memory(Bi-LSTM)network model was proposed.Firstly,the HI curve of high-speed shaft bearing was constructed by using monotonicity analysis and KPCA,Then the health state of high-speed shaft bearing was predicted by using Bi-LSTM network on the basis of the constructed HI curve.The HI curve constructed by this method is proved to have good monotonicity and can be used for more accurate health assessment and state prediction of high-speed shaft bearings.
作者
李振恩
张新燕
胡威
谢丽蓉
Li Zhen’en;Zhang Xinyan;Hu Wei;Xie Lirong(College of Electrical Engineering,Xinjiang University,Urumqi 830047,China)
出处
《太阳能学报》
EI
CAS
CSCD
北大核心
2021年第10期290-297,共8页
Acta Energiae Solaris Sinica
基金
国家自然科学基金(51667018)。
关键词
风电机组
轴承
预测
评估
健康指数
核主元分析
双向长短期记忆
wind turbines
bearings
prediction
assessment
health index
kernel principal component analysis(KPCA)
bidirectional long shot term memory(Bi-LSTM)
