摘要
轴承故障诊断一直是机械状态监测中的重点及难点之一。近年来,卷积神经网络(convolutional neural network,CNN)在机器状态监测中得到了广泛的应用,其中在轴承故障诊断领域中也具有很大突破。但由于CNN提取一维特征时由于自身网络特征而产生的一些缺点,导致传统CNN对故障轴承原始振动信号进行处理所得的模型准确率较低且泛化性较差。基于上述问题,改进了传统CNN中的卷积层,提出了改进的CNN网络,将原始轴承振动信号作为输入,提取了信号的深层特征,将该特征输入到分类器中进行轴承故障分类。同时利用内蒙古科技大学轴承振动信号数据集及西储大学轴承振动信号数据集对提出网络进行了有效性论证及泛化性论证,实验表明:相较于传统CNN网络,该网络更具有稳定性,且更具有泛化性。
Bearing fault diagnosis is one of the key and difficult problems in mechanical condition monitoring.In recent years,convolutional neural network(CNN)has been widely applied in machine state monitoring,among which it has made great breakthroughs in the field of bearing fault diagnosis.However,due to the shortcomings of CNN in extracting one-dimensional features due to its own network features,the model obtained by traditional CNN in processing the original vibration signals of fault bearings has low accuracy and poor generalization.Based on the above problems,the convolutional layer in the traditional CNN was improved,an improved CNN network was proposed,the original bearing vibration signal as the input was took,the deep feature of the signal was extracted,and the feature into the classifier for bearing fault classification was input.At the same time,the validity and generalization of the proposed network were demonstrated by using the bearing vibration signal data set of Inner Mongolia university of science and technology and the bearing vibration signal data set of western reserve university.
作者
杨兰柱
刘文广
Yang Lanzhu;Liu Wenguang(Engineering and Training Center,Inner Mongolia University of Science and Technology,Baotou,Inner Mongolia 014010,China;Inner Mongolia University of Science and Technology,Baotou,Inner Mongolia 014010,China)
出处
《机电工程技术》
2020年第8期11-13,共3页
Mechanical & Electrical Engineering Technology
关键词
卷积神经网络
改进的CNN网络
模式识别
故障诊断
convolutional neural network
improved CNN network
pattern recognition
fault diagnosis
