Looseness diagnosis method for connecting bolt of fan foundation based on sensitive mixed-domain features of excitation-response and manifold learning

详细信息    查看全文

• 作者：Renxiang Chen^a ; ^bAuthor Vitae ; Siyang Chen^aAuthor Vitae ; Lixia Yang^a ; ^cAuthor Vitae ; Jiaxu Wang^b ; ^{manlouyue@126.com}Author Vitae ; Xiangyang Xu^aAuthor Vitae ; Tianhong Luo^aAuthor Vitae
• 关键词：Looseness diagnosis ; Connecting bolt ; Excitation-response ; Sensitive feature ; Manifold learning
• 刊名：Neurocomputing
• 出版年：2017
• 出版时间：5 January 2017
• 年：2017
• 卷：219
• 期：Complete
• 页码：376-388
• 全文大小：2103 K
• 卷排序：219

文摘

Propose a method based on excitation-response to characterize the bolt looseness. Use 20 mixed-domain features to construct high-dimensional looseness feature set. Propose an algorithm of looseness sensitivity index to select the sensitive feature. Introduce ONPE and WKNNC for dimension reduction and looseness recognition. Verify the effectiveness proposed method by connecting bolt looseness diagnosis.