【摘要】 提出了一种数据处理方法,对滚动轴承振动数据周期进行自动划分;将数据按照划分的数据长度传入一维多尺度卷积神经网络(ODM-CNN)进行滚动轴承的故障诊断。所提出的数据处理方法可以将每一次传入神经网络的数据划分出合适的周期。通过这种方式滚动轴承故障诊断的准确率达到99.5%以上,并且有效提高了网络的反应速率。更多还原

【Abstract】 Proposed a data processing method, which automatically divides the rolling bearing vibration data cycle;transmitted the data into one-dimensional multiscale convolutional neural network(ODM-CNN)according to the divided data length for rolling bearing fault diagnosis. The data processing method proposed can divide the appropriate period for the data each incoming neural network. In this way, the accuracy of rolling bearing fault diagnosis is more than 99.5%, and effectively improves the reaction rate of the network.更多还原