Abnormal Detection and Fault Diagnosis Method of Bearing Based on Deep Convolutional Generative Adversarial Network

Abnormal Detection and Fault Diagnosis Method of Bearing Based on Deep Convolutional Generative Adversarial Network

연구 분야: Artificial Intelligence

학회: 2023 IEEE 11th Joint International Information Technology and Artificial Intelligence Conference (ITAIC)

The breakdowns of Rolling bearings have a large impact range and then cause high maintenance costs. To solve this problem, this paper studies a small-sample bearing anomaly detection and fault diagnosis method based on deep convolutional generative adversarial network. This method constructs a DC-GANomaly network model based on attention mechanism by fusing DCGAN and GANomaly networks and introducing attention mechanisms. In this paper, the bearing fault dataset of Western Reserve University is used to verify the proposed method, and the performance of the model is discussed in comparative experiments.