面向自动化试验产线的旋转机械故障诊断方法

Rotating Machinery Fault Diagnosis Methods for Automated Test Production Lines

中文关键词:  产线  旋转机械  故障诊断  DCCNN  BiLSTM

英文关键词:production line  rotating machinery  fault diagnosis  DCCNN  BiLSTM

中文摘要:

      近年来，随着半导体行业的不断发展，自动化试验产线随之出现，以应对测试任务骤增的现象。旋转机械作为自动化试验产线的核心设备，不可避免地会遇到各种故障，在生产运行中可能导致较大的损失。为保障自动化试验产线的正常运行，开发可行的旋转机械故障诊断方法具有重要意义。针对航天科工防御技术研究试验中心自行建立的微电子器件自动化试验产线中常见的旋转机械，提出了一种基于双通道卷积神经网络和双向长短期记忆网络（Dual-Channel Convolutional Neural Network and Bidirectional Long Short-Term Memory Network，DCCNN-BiLSTM）的旋转机械故障诊断方法，该方法不仅克服了传统算法中对信号数据信息利用不充分的问题，而且可以对原始信号进行特征自动提取和故障诊断，即无须对输入的原始信号进行任何人工处理便可直接完成故障诊断，更加简单高效。

英文摘要:

      In recent years,with the continuous development of the semiconductor industry,the automated test production lines appeared to cope with the phenomenon of the sudden increase in testing tasks.Rotating machinery as the core equipment of the automated test production lines,will inevitably encounter a variety of failures,which may lead to significant losses in production operation.In order to ensure the normal operation of the automated test production lines,it is of great significance to develop a feasible fault diagnosis method for rotating machinery.A rotating machinery fault diagnosis method based on dual-channel convolutional neural network and bidirectional long and short-term memory network (DCCNN-BiLSTM) is proposed for the common rotating machinery in the microelectronic device automated test production lines established by China Aerospace Science & Industry Corp Defense Technology R&T Center,which not only overcomes the problem of underutilizing the information of the signal data in traditional algorithms but also can automatically extract the features and diagnose faults from the original signals.That is,without any manual processing on the input original signals,the fault diagnosis can be accomplished directly,which makes it easier and more efficient.