Intelligent Identification of Rotor Axis Trajectory Based on Anti-Grayscale Preprocessing and Convolutional Neural Networks

Different shapes of rotor axis trajectory can reflect different fault states of the rotating machinery. However, the current data processing methods can not extract the information characteristics of rotor axis trajectory as a symptom of intelligent fault diagnosis of the rotating equipment. In this paper, the image recognition method is used to transform the problem of vibration signal analysis in the orthogonal direction into the problem of pattern recognition of the two-dimensional image. The anti-grayscale preprocessing method can effectively prevent the image from losing contour information after the max pooling operation. The convolutional neural network is used to extract the local and global topological features of the rotor axis trajectory images to eliminate the influence of plane position on recognition. Finally, the information description of different rotor axis trajectory shapes is obtained, which is used as the feature of intelligent fault diagnosis of rotating equipment. The experimental results show that the rotor axis trajectory images pretreated by the anti-grayscale preprocessing method have more advantages in the process of training the convolutional neural network. Compared with the traditional methods of recognizing the rotor axis trajectory, the intelligent recognition method based on the convolutional neural network has higher accuracy and better robustness.