A Novel Fault Diagnosis Strategy for Diaphragm Pumps Based on Signal Demodulation and PCA-ResNet

Author:

Meng Fanguang12,Shi Zhiguo1ORCID,Song Yongxing34

Affiliation:

1. College of Information Science and Electronic Engineering, Zhejiang University, Hangzhou 310013, China

2. Zhejiang JingLiFang Digital Technology Group Co., Ltd., Hangzhou 310012, China

3. School of Thermal Engineering, Shandong Jianzhu University, Jinan 250101, China

4. State Key Laboratory of Compressor Technology (Compressor Technology Laboratory of Anhui Province), Hefei 230031, China

Abstract

The efficient and accurate identification of diaphragm pump faults is crucial for ensuring smooth system operation and reducing energy consumption. The structure of diaphragm pumps is complex and using traditional fault diagnosis strategies to extract typical fault characteristics is difficult, facing the risk of model overfitting and high diagnostic costs. In response to the shortcomings of traditional methods, this study innovatively combines signal demodulation methods with residual networks (ResNet) to propose an efficient fault diagnosis strategy for diaphragm pumps. By using a demodulation method based on principal component analysis (PCA), the vibration signal demodulation spectrum of the fault condition is obtained, the typical fault characteristics of the diaphragm pump are accurately extracted, and the sample features are enhanced, reducing the cost of fault diagnosis. Afterward, the PCA-ResNet model is applied to the fault diagnosis of diaphragm pumps. A reasonable model structure and advanced residual block design can effectively reduce the risk of model overfitting and improve the accuracy of fault diagnosis. Compared with the visual geometry group (VGG) 16, VGG19, ResNet50, and autoencoder models, the proposed model has improved accuracy by 35.89%, 80.27%, 2.72%, and 6.12%. Simultaneously, it has higher operational efficiency and lower loss rate, solving the problem of diagnostic lag in practical engineering. Finally, a model optimization strategy is proposed through model evaluation metrics and testing. The reasonable parameter range of the model is obtained, providing a reference and guarantee for further optimization of the model.

Funder

Natural Science Foundation of Shandong Province

Open Foundation of State Key Laboratory of Compressor Technology

Publisher

MDPI AG

Cited by 2 articles. 订阅此论文施引文献 订阅此论文施引文献,注册后可以免费订阅5篇论文的施引文献,订阅后可以查看论文全部施引文献

同舟云学术

1.学者识别学者识别

2.学术分析学术分析

3.人才评估人才评估

"同舟云学术"是以全球学者为主线,采集、加工和组织学术论文而形成的新型学术文献查询和分析系统,可以对全球学者进行文献检索和人才价值评估。用户可以通过关注某些学科领域的顶尖人物而持续追踪该领域的学科进展和研究前沿。经过近期的数据扩容,当前同舟云学术共收录了国内外主流学术期刊6万余种,收集的期刊论文及会议论文总量共计约1.5亿篇,并以每天添加12000余篇中外论文的速度递增。我们也可以为用户提供个性化、定制化的学者数据。欢迎来电咨询!咨询电话:010-8811{复制后删除}0370

www.globalauthorid.com

TOP

Copyright © 2019-2024 北京同舟云网络信息技术有限公司
京公网安备11010802033243号  京ICP备18003416号-3