机械研究与应用
機械研究與應用
궤계연구여응용
Mechanical Research & Application
2015年
4期
177-180
,共4页
提升小波%流形学习%抽油机系统%故障特征提取
提升小波%流形學習%抽油機繫統%故障特徵提取
제승소파%류형학습%추유궤계통%고장특정제취
lifting wavelet%manifold learning%pumping unit system%fault feature extraction
针对常用的信号处理算法在早期故障、微弱故障、复合故障等信号的处理方面存在的不足,提出了小波变换的方法。重点介绍了提升小波和流形学习LE算法相结合提取系统故障特征。解决了LE对噪声鲁棒性较差的问题,增强了流形学习在信号处理中的优越性。经水基动力无杆抽油机抽油机系统故障模拟试验台验证,该方法能准确提取抽油机故障特征,对抽油机故障进行分类与识别,为后续分析奠定了基础。
針對常用的信號處理算法在早期故障、微弱故障、複閤故障等信號的處理方麵存在的不足,提齣瞭小波變換的方法。重點介紹瞭提升小波和流形學習LE算法相結閤提取繫統故障特徵。解決瞭LE對譟聲魯棒性較差的問題,增彊瞭流形學習在信號處理中的優越性。經水基動力無桿抽油機抽油機繫統故障模擬試驗檯驗證,該方法能準確提取抽油機故障特徵,對抽油機故障進行分類與識彆,為後續分析奠定瞭基礎。
침대상용적신호처리산법재조기고장、미약고장、복합고장등신호적처리방면존재적불족,제출료소파변환적방법。중점개소료제승소파화류형학습LE산법상결합제취계통고장특정。해결료LE대조성로봉성교차적문제,증강료류형학습재신호처리중적우월성。경수기동력무간추유궤추유궤계통고장모의시험태험증,해방법능준학제취추유궤고장특정,대추유궤고장진행분류여식별,위후속분석전정료기출。
In view of deficiencies that the common signal processing algorithm in the processing of signals such as the early failure, weak fault and compound fault signal, the method of wavelet transform is put forward, focused on the combination of lifting wavelet transform and LE manifold learning algorithm to extract fault feature. The poor noise robustness problem of LE is solved, and the superiority of manifold learning in signal processing is enhanced. Through verification of the fault simulation experiment platform of water-powered rod-less pumping unit, it is proved that the method can accurately extract the fault fea-tures of pumping unit, classify and distinguish the fault of pumping system, and to provide the basis for subsequent analysis.