摘要
为准确、可靠、有效地进行起重机泵健康评估,通过对多台不同使用年限的起重机进行试验,获取起重机实际运行时的泵信号,采用深度自编码模型(DAE)与马氏距离(MD)结合的方式,揭示泵性能退化和评估健康状况,对比统计特征、EMD、MLP、CNN等特征学习方法,表明该方法具有良好的性能。
In order to evaluate the health status of the crane pump accurately,reliably and effectively,the pump of the crane signals during actual operation of the crane are obtained by testing several cranes with different service years.The combination of the deep auto-encoding model(DAE)and the Mahalanobis distance(MD)is used to reveal the pump performance degradation and evaluate the health status.By comparing the statistical feature,EMD,MLP,CNN and other feature learning methods,this method has been shown to have good performance.
作者
屈斌
李彬
Qu Bin;Li Bin(Yangtze Three Gorges Navigation Administration;Weite Technologies Co.,Ltd)
出处
《港口装卸》
2021年第3期49-51,共3页
Port Operation
关键词
健康评估
起重机泵
深度自编码模型
度量学习
health assessment
crane pump
deep auto-encoding model
metric learning