摘要
针对岸桥机构减速箱振动信号局部扰动特征学习问题的复杂性,提出一种新型故障诊断模型。首先利用整体经验模态分解(Ensemble Empirical Mode Decomposition,EEMD)变换对振动信号进行瞬态时频分析,获取典型参数的边际谱等特征,形成深度学习的特征向量。在此基础上,提出一种基于深度收缩自编码网络-模糊支持向量机的起重机械故障状态识别模型,并与深度稀疏自编码网络-模糊支持向量机模型等其它模型进行比较。实验结果显示,针对起重机械故障状态识别问题,所建立的新深度学习模型有很好的识别能力,识别准确率可达95.6%。
Addressing the complexity of learning the local disturbance characteristics of the vibration signal of the quay crane mechanism,a new type of fault diagnosis model is proposed.Firstly,the Ensemble Empirical Mode Decomposition(EEMD)transformation is used to analyze the instantaneous frequency of typical vibration signals,obtain their marginal spectrum and other characteristics,and form the feature vector of deep learning.On this basis,a crane fault diagnosis model based on deep contractive auto-encoder network-fuzzy support vector machine is proposed and compared with the traditional methods.The experimental results show that the new deep learning model is good at classifying vibration signals of quay crane,and the classification accuracy rate can reach 95.6%.
作者
钱尼君
李勇
彭献勇
吴奇
QIAN Nijun;LI Yong;PENG Xianyong;WU Qi(Jiangxi Special Equipment Inspection and Testing Institute,Nanchang 330096,China;Tengyi Data(Shanghai)Technology Co.,Ltd.,Shanghai 200240,China;School of Electronic Information and Electrical Engineering,Shanghai Jiaotong University,Shanghai 200240,China)
出处
《机械设计与研究》
CSCD
北大核心
2021年第5期78-82,共5页
Machine Design And Research
基金
国家自然科学基金(61671293)
江西省重点研发计划(20192BBE50065)资助项目。
