零故障样本下小波知识驱动的工业机器人故障检测  

Wavelet knowledge-driven mechanical equipment fault detection with zero-fault samples

作　　者：黎国强 魏美容 吴德烽[1] 吴军[3] 段超群 Li Guoqiang;Wei Meirong;Wu Defeng;Wu Jun;Duan Chaoqun(School of Marine Engineering,Jimei University,Xiamen 361021,China;Navigation Institute,Jimei University,Xiamen 361021,China;School of Naval Architecture and Ocean Engineering,Huazhong University of Science and Technology,Wuhan 430074,China;School of Mechatronic Engineering and Automation,Shanghai University,Shanghai 200444,China)

机构地区：[1]集美大学轮机工程学院,厦门361021 [2]集美大学航海学院,厦门361021 [3]华中科技大学船舶与海洋工程学院,武汉430074 [4]上海大学机电工程与自动化学院,上海200444

出　　处：《仪器仪表学报》2024年第9期166-176,共11页Chinese Journal of Scientific Instrument

基　　金：国家自然科学基金面上项目(52375106);厦门市自然科学基金项目(3502Z202471042)资助。

摘　　要：针对零故障样本问题,现有方法大多从迁移学习、样本生成等开展研究,然而该类方法依赖相近故障样本,难以保证训练样本与真实故障信号在数据分布上保持对齐,导致模型泛化性不足。针对上述问题,提出了基于连续小波变换知识库和ViT网络的故障检测方法。采用了多种母小波函数构建连续小波变换知识库,从不同时-频角度对机械装备监测数据进行分析;设计了一种基于多模态时-频特征的对比损失函数,实现了ViT的有效训练;开发了基于余弦相似性分析的故障检测算法,检测机械装备各类异常状态。使用工业机器人实验平台对方法进行验证。结果表明,所提方法能够在零故障样本下构建高性能的特征提取网络,并能对各类故障状态进行准确检测。For the zero-fault sample problem of mechanical equipment,transfer learning-based and data generation-based methods have attracted much attention.However,the methods usually require the support of similar fault samples,which makes it difficult to ensure training data is aligned with the real-world fault sample of mechanical equipment in data distribution.The generalization is insufficient in applications.To address the aforementioned problems,a novel fault detection(FD)method is proposed based on a designed new loss function and vision transformer(ViT).First,the continuous wavelet transform knowledgebase is established by combining three different mother wavelet functions,which are used to analyze the monitored signals of mechanical equipment from different time-frequency perspectives.Secondly,a new contrastive loss function is designed based on different time-frequency features and the cosine similarity analysis to effectively optimize the parameters of a constructed ViT.Finally,a fault detection algorithm is proposed to parse the real-time monitored signals of mechanical equipment to achieve the FD.The proposed FD method is evaluated by the industrial robot test rig.The results show that a deep network-based feature encoder with high-performance feature extraction can be established with zero-fault samples,and accurate fault detection for different fault conditions can also be realized.For the zero-fault sample problem of mechanical equipment,transfer learning-based and data generation-based methods have attracted much attention,however the methods usually require the support of similar fault samples,and which are difficult to ensure training data aligned with the real-world fault sample of mechanical equipment in data distribution,and their generalization are insufficient in applications.To solve above problem,a novel fault detection(FD)method is proposed based on a designed new loss function and vision transformer(ViT).First,a continuous wavelet transform knowledgebase is established by combining three differ

关 键 词：零故障样本 连续小波变换知识库 对比损失函数 故障检测 

分 类 号：TH17[机械工程—机械制造及自动化]