A Fault Diagnosis Method Based on a Rainbow Recursive Plot and Deep Convolutional Neural Networks

In previous deep learning-based fault diagnosis methods for rotating machinery, the method of directly feeding one-dimensional data into convolutional neural networks can lead to loss important features. To address problem, a novel machinery model based on rainbow recursive plot (RRP) is proposed. Our main innovation and contributions are: First, RRP proposed convert vibration signal from two-dimensional color image, facilitating capturing more significant information. Second, new CNN LeNet-5 devised, which extracts feature that describes substantial information converted thus performing recognition accurately. The public rolling bearing datasets online platform are adopted verify performance. Experiments show improve accurate rate 97.86%. More importantly, experiment self-made demonstrates our approach achieves best comprehensive performance in terms speed accuracy compared mainstream algorithms.