A New Method Based on Encoding Data Probability Density and Convolutional Neural Network for Rotating Machinery Fault Diagnosis

In order to apply the advantages of image recognition for fault diagnosis using convolutional neural network (CNN), it is necessary convert one-dimensional (1D) signal data into two-dimensional (2D) images. Traditional signal-based conversion methods face challenges complex feature extraction process, high dependence on expert knowledge and poor repeatability, which hinder timely accurate diagnosis. Therefore, this paper proposes a method rotating machinery in virtue Data Probability Density-Gram Angle Field-Convolutional Neural Network (DPD-GAF-CNN). DPD-GAF first computes DPD 1D time series through parameter-free statistics, then encodes 2D that directly reflects mean standard deviation probability distribution. Besides simplified transformation no artificially designed features are required like original GAF encode process After that, CNN based LeNet-5 used achieve high-precision classification. The proposed verified compared with other existing intelligent experimental generated by planetary gearbox test bench various faulty conditions bearing set Case Western Reserve University. results show presented can effectively improve accuracy stability classification several datasets up 99.9%.