Discriminant Feature Extraction of Motor Current Signal Analysis and Vibration For Centrifugal Pump Fault Detection

Asma’ul Husna, K. Indriawati, B. L. Widjiantoro
{"title":"Discriminant Feature Extraction of Motor Current Signal Analysis and Vibration For Centrifugal Pump Fault Detection","authors":"Asma’ul Husna, K. Indriawati, B. L. Widjiantoro","doi":"10.1109/ICA52848.2021.9625679","DOIUrl":null,"url":null,"abstract":"The monitoring condition of the centrifugal pump is closely related to fault detection and diagnosis. It usually uses the vibration signals. However, under certain conditions it is not possible to install the accelerometer on the machine due to certain conditions and environments. Current signals can be used to replace vibration signals. This method is called motor current signature analysis (MCSA). The raw signal of the current and current spectrum in frequency domain can be used for fault detection. The statistical features of the current raw signal contain information on the signal characteristics. However, these raw features are not sensitive enough to weak fault symptoms or are not suitable for severe faults, thus can affect fault detection and classification accuracy. To overcome this problem, discriminant feature extraction is carried out for fault detection in centrifugal pumps (CP). Discriminant features are divided into four phases. In the first phase, a healthy pump signal is selected. In the second phase, the healthy condition signal is cross-correlated with the centrifugal pump current signal in several fault classes and the result of the extraction from the cross correlation is a new feature set. In the third phase, the raw statistical features in the time, frequency and time-frequency domains are extracted from both healthy current signals and CP current signals of different classes. In the last phase, wavelet packet transform (WPT) energy is extracted from the current signals. The result of these features will be combined into a discriminant feature pool. The pool discriminant feature will be used as input in making a classifier for the centrifugal pump fault detection system. This study also used motor bearing speed data for comparison. The main topic of this paper is to design a fault detection system for centrifugal pumps using current signals. Based on the performance test using precision, error rate, and recall. The motor bearing speed vibration signal has better performance than the CP fault detection classification with the current signal. However, there is only a slight difference between the two. From this research, the current signal and motor bearing speed vibration signal can detect fault to the centrifugal pump well.","PeriodicalId":287945,"journal":{"name":"2021 International Conference on Instrumentation, Control, and Automation (ICA)","volume":"85 1","pages":"0"},"PeriodicalIF":0.0000,"publicationDate":"2021-08-25","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"2","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"2021 International Conference on Instrumentation, Control, and Automation (ICA)","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1109/ICA52848.2021.9625679","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 2

Abstract

The monitoring condition of the centrifugal pump is closely related to fault detection and diagnosis. It usually uses the vibration signals. However, under certain conditions it is not possible to install the accelerometer on the machine due to certain conditions and environments. Current signals can be used to replace vibration signals. This method is called motor current signature analysis (MCSA). The raw signal of the current and current spectrum in frequency domain can be used for fault detection. The statistical features of the current raw signal contain information on the signal characteristics. However, these raw features are not sensitive enough to weak fault symptoms or are not suitable for severe faults, thus can affect fault detection and classification accuracy. To overcome this problem, discriminant feature extraction is carried out for fault detection in centrifugal pumps (CP). Discriminant features are divided into four phases. In the first phase, a healthy pump signal is selected. In the second phase, the healthy condition signal is cross-correlated with the centrifugal pump current signal in several fault classes and the result of the extraction from the cross correlation is a new feature set. In the third phase, the raw statistical features in the time, frequency and time-frequency domains are extracted from both healthy current signals and CP current signals of different classes. In the last phase, wavelet packet transform (WPT) energy is extracted from the current signals. The result of these features will be combined into a discriminant feature pool. The pool discriminant feature will be used as input in making a classifier for the centrifugal pump fault detection system. This study also used motor bearing speed data for comparison. The main topic of this paper is to design a fault detection system for centrifugal pumps using current signals. Based on the performance test using precision, error rate, and recall. The motor bearing speed vibration signal has better performance than the CP fault detection classification with the current signal. However, there is only a slight difference between the two. From this research, the current signal and motor bearing speed vibration signal can detect fault to the centrifugal pump well.
用于离心泵故障检测的电机电流信号分析与振动判别特征提取
离心泵的监测状况与故障的检测和诊断密切相关。它通常使用振动信号。但是,在某些条件下,由于某些条件和环境,不可能在机器上安装加速度计。电流信号可用来代替振动信号。这种方法被称为电机电流特征分析(MCSA)。电流和电流谱的原始信号在频域内可用于故障检测。当前原始信号的统计特征包含了信号特性的信息。然而,这些原始特征对较弱的故障症状不够敏感,或者不适用于较严重的故障,从而影响故障检测和分类的准确性。为了克服这一问题,采用判别特征提取方法对离心泵进行故障检测。判别特征分为四个阶段。在第一阶段,选择一个健康的泵信号。在第二阶段,将健康状态信号与离心泵电流信号在若干故障类别中相互关联,从相互关联中提取的结果是一个新的特征集。第三阶段分别从不同类别的健康电流信号和CP电流信号中提取时域、频域和时频域的原始统计特征。最后一阶段,从电流信号中提取小波包变换能量。这些特征的结果将被组合成一个判别特征池。将池判别特征作为输入,制作离心泵故障检测系统的分类器。本研究还使用电机轴承转速数据进行比较。本文的主要课题是设计一个利用电流信号检测离心泵故障的系统。基于使用精度、错误率和召回率的性能测试。电机轴承转速振动信号与电流信号相比,具有更好的故障检测分类性能。然而,两者之间只有细微的差别。研究表明,电流信号和电机轴承转速振动信号能较好地检测出离心泵的故障。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
自引率
0.00%
发文量
0
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:604180095
Book学术官方微信