摘要
文中针对泥泵SCADA数据工况混杂、含有“脏数据”等问题,采用聚类算法、双向清洗等方法进行处理.建立泥泵健康状态评估指标体系,利用EEMD-NARX神经网络进行趋势预测.以单位时间内的维修费用来作为评价准则,通过两种维修策略对比分析,验证了预测性维修策略的优势.
Aiming at the problems of mixed working conditions and“dirty data”in SCADA data of sludge pump,clustering algorithm and two-way cleaning were adopted for processing.The health evaluation index system of mud pump was established,and the trend was predicted by EEMD-NARX neural network.Taking the maintenance cost per unit time as the evaluation criterion,the advantages of predictive maintenance strategy were verified through the comparative analysis of two maintenance strategies.
作者
邓义斌
黄龙归
赵六军
DENG Yibin;HUANG Longgui;ZHAO Liujun(School of Naval Architecture,Ocean and Energy Power Engineering,Wuhan University of Technology,Wuhan 430063,China;Changjiang Wuhan Waterway Engineeing Bureau,Wuhan 430014,China)
出处
《武汉理工大学学报（交通科学与工程版）》
2023年第5期866-869,875,共5页
Journal of Wuhan University of Technology（Transportation Science & Engineering）
基金
工业互联网标识解析二级节点(运输行业应用服务平台)项目(TC190A3X8-7)。
关键词
预测性维护技术
泥泵
趋势预测
数据处理
predictive maintenance technology
mud pump
trend prediction
data processing
