基于非线性自回归卷积的电梯门机故障诊断方法  

A fault diagnosis method for elevator door machine based on nonlinear autoregressive convolutions

作　　者：肖刚[1,2] 张皓鑫 王亚明 徐雪松 张元鸣[1] 程振波[1] XIAO Gang;ZHANG Haoxin;WANG Yaming;XU Xuesong;ZHANG Yuanming;CHENG Zhengbo(College of Computer Science and Technology,Zhejiang University of Technology,Hangzhou 310023,China;College of Mechanical Engineering,Zhejiang University of Technology,Hangzhou 310023,China)

机构地区：[1]浙江工业大学计算机科学与技术学院,浙江杭州310023 [2]浙江工业大学机械工程学院,浙江杭州310023

出　　处：《浙江工业大学学报》2024年第6期601-610,共10页Journal of Zhejiang University of Technology

基　　金：国家自然科学基金青年项目(52205291);国家自然科学基金面上项目(61976193);浙江省“尖兵”研发攻关计划项目(2023C01022);浙江省“领雁”研发攻关计划项目(2023C01215)。

摘　　要：经由振动传感器获取的电梯门机运行信号是一类典型多属性时间序列,根据这类多属性时间序列进行故障分类,需要能同时实现序列的空间与时间特征编码。为此,提出了一种基于非线性向量自回归卷积神经网络模型,用于多属性时间序列的空时特征编码。该模型首先通过非线性向量自回归实现序列时间维度的特征提取;然后利用卷积和池化提取序列的空间特征;最后经由全连接层实现对时间序列的分类。通过振动传感器获取电梯门机的多属性时间序列信号,利用模型实现了基于上述时间序列信号的电梯门机故障诊断,从而验证了模型在多属性空时信号特征提取方面具有的优势。The operating signal of the elevator door machine obtained through vibration sensors is a typical type of multi-attribute time series.To classify faults based on such b multi-attribute time series,it is necessary to simultaneously encode the spatial and temporal features of the sequence.Therefore,a nonlinear vector autoregressive convolutional neural network model is proposed for spatiotemporal feature encoding of multi-attribute time series.The model first extracts features from the time dimension of the sequence through nonlinear vector auto-regression.Then,the convolution and pooling are used to extract spatial features of the sequence;Finally,the classification of time series can be achieved through the fully connected layer.The multi-attribute time series signals of the elevator door machine are obtained through vibration sensors,and the model is used to achieve fault diagnosis of the elevator door machine based on the above time series signals.It verifies the advantages of the model in feature extraction of multi-attribute space-time signals.

关 键 词：电梯门机 故障诊断 多属性时间序列 非线性向量自回归 卷积神经网络 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]