基于SSA-VMD和DPCNN-BiGRU的微型电机异音检测  

Micro-Motor Abnormal Sound Detection Based on SSA-VMD and DPCNN-BiGRU

作　　者：候嫣茹 章艺[1] 符栋梁 郭政彤 徐子钦 钱之纯 HOU Yanru;ZHANG Yi;FU Dongliang;GUO Zhengtong;XU Ziqin;QIAN Zhichun(Shanghai Marine Equipment Research Institute,Shanghai 200031,China)

机构地区：[1]上海船舶设备研究所,上海200031

出　　处：《船舶工程》2024年第12期118-129,共12页Ship Engineering

摘　　要：针对某船用微型电机在出厂异音检测环节中存在人工检测准确率低、主观性强等问题,提出了一种基于优化的变分模态分解(VMD)和双路径卷积神经网络(DPCNN)联合双向门控循环单元(BiG RU)的异音检测方法,首先采用麻雀搜索算法(SSA)优化VMD自适应分解信号进行降噪,其次利用DPCNN和高效通道注意力(ECA)机制提取特征信息,最后通过BiGRU挖掘信号中的时间相关规律,实现对微型电机异音的准确识别。用测试集进行试验,结果表明:该方法能够有效检测微型电机的异音,准确率高达97.68%,与其他深度学习模型相比具有明显的特征提取和检测性能优势。In order to overcome the problems of low accuracy and strong subjectivity of manual detection in the abnormal sound detection process of micro motor factory,an abnormal sound detection method based on optimized variational mode decomposition(VMD)and dual-path convolutional neural network(DPCNN)combined with bidirectional gated recurrent unit(BiGRU)is proposed.Firstly,the sparrow search algorithm(SSA)is used to optimize the VMD adaptive decomposition signal for noise reduction,then the DPCNN and efficient channel attention(ECA)mechanism are used to extract the feature information,and finally the time correlation information in the signal is mined by BiGRU to achieve accurate recognition of the abnormal micro motor sound.Experiments on the test dataset show that the proposed method can effectively detect the abnormal sound of the micro motor with an accuracy of up to 97.68%,which has obvious advantages in feature extraction and detection performance compared with other deep learning models.

关 键 词：异音检测 变分模态分解 高效通道注意力 双向门控循环单元 

分 类 号：TH165.3[机械工程—机械制造及自动化]