中文摘要：

提出一种基于小波时频图和卷积神经网络（CNN）的滚动轴承智能故障诊断方法。对滚动轴承的振动信号进行连续小波变换（CWT），得到时频图，并以灰度图的形式显示，再将时频图压缩至适当的大小；将压缩后的时频图作为特征图输入，建立CNN分类器模型，以实现滚动轴承的智能故障诊断。基于人工轴承故障数据集进行了实验研究，同时从结构参数和训练参数两方面对网络的性能进行了优化改进。结果表明，该方法能有效识别滚动轴承的故障类型，改进的CNN具有较强的泛化能力、特征提取和识别能力。

英文摘要：

An approach to intelligent fault diagnosis of rolling bearing using wavelet time-frequency representations and convolution neural network（CNN） was proposed. It used continuous wavelet transform （CWT） to analyze vibration signals of rolling bearing and get time-frequency representations in grey-scale. Then, the time- frequency representations were compressed to the appropriate size. After that, all the compressed time-frequency representations were as input feature maps, and CNN was developed to identify the faults of rolling bearing. Tests for the proposed method were accomplished based on artificial bearing fault data sets, and the performance of the network was optimized from structural parameters and training parameters. The experimental results indicates that this method could effectively identify the fault patterns of rolling bearing, and the improved CNN has strong ability of generalization, feature extraction and recognition.