Intelligent Fault Diagnosis of Gearbox in Rotating Machinery Using Adversarial Neural Networks under Variable Speeds

This research presents an intelligent fault diagnosis method for gearbox systems in rotating machinery using Generative Adversarial Networks (GANs). Due to variable speeds and the scarcity of reliable data in real-world conditions, accurate fault detection faces significant challenges. While GANs can generate realistic synthetic data, this study leverages adversarial training, aiming to make the diagnostic system resilient to speed variations. By combining adversarial learning with real data, diagnostic models can better recognize various fault patterns and improve detection accuracy. The primary goal of this study is to enhance the precision and reliability of gearbox fault diagnosis under different operational conditions using GANs. With deep learning techniques, the proposed approach can identify various gearbox faults even under limited and fluctuating data conditions. Experimental results demonstrate that this approach improves detection performance and fault classification accuracy compared to traditional methods.