摘要
提出了一种基于小波阈值降噪算法和支持向量机(SVM)模型的疏水泵基础平台螺栓松动故障诊断方法。获得疏水泵基础平台的振动时域信号,运用小波阈值降噪算法去除噪声和无关成分的干扰,对信号进行了降噪和重构,提取了降噪信号的4类时域敏感特征作为特征向量集,将其导入SVM模型进行故障识别训练与测试。结果表明:该方法可以获得振动信号特征值,识别螺栓松动故障特征,正确判断故障类型,对确保疏水泵安全可靠运行具有重要意义。
A fault diagnosis method based on wavelet threshold noise reduction algorithm and support vector machine(SVM)model was proposed for the loose bolt fault in the drainage pump’s foundation platform,including obtaining vibration time-domain signals of the pump foundation platform.Through applying the wavelet threshold noise reduction algorithm to remove the interference of noise and irrelevant components,the signals were denoised and reconstructed and the four types of time-domain sensitive features of the denoised signals were extracted as the set of feature vectors and then imported into the SVM model for the fault identification training and testing.The results show that,this method can obtain vibration signal eigenvalues,identify the loose bolt’s fault characteristics and determine fault types.It has great significance in ensuring both safe and reliable operation of the drainage pump.
作者
万益恒
薛金山
张大勇
韩学杰
吕井
董省身
李海生
WAN Yi-heng;XUE Jin-shan;ZHANG Da-yong;HAN Xue-jie;LV Jing;DONG Xing-shen;LI Hai-sheng(Yangjiang Nuclear Power Co.,Ltd.;Hangzhou Anmaisheng Intelligent Technology Co.,Ltd.;School of Chemical Engineering and Technology,China University of Mining and Technology)
出处
《化工机械》
2025年第4期666-673,共8页
Chemical Engineering & Machinery
基金
先进反应堆工程与安全教育部重点实验室开放基金(批准号:ARES-2024-05)资助的课题
国家自然科学基金(批准号:51105362)资助的课题
阳江核电有限公司科技项目(批准号:3100172380)资助的课题。
关键词
疏水泵
螺栓松动
振动信号
降噪
故障诊断
drainage pump
loose bolt
vibration signal
denoise
fault diagnosis
