SAE-WGAN Based Fault Diagnosis Method for Rolling Mills System

The performance and condition of the mill hydraulic automatic gauge control (AGC) hydraulic system has a direct impact on the final roll quality. This method combines stacked auto-encoders (SAE)and generative adversarial network (GAN), using fault signals collected from sensors, in depth training and simulation of fault data, and enhancing the accuracy and efficiency of fault diagnosis by generative adversarial network to optimize the model's learning and realistic sample generation capability. The experimental results demonstrate that this method combining stacked auto-encoders and GAN is not only applicable to the complex AGC system of rolling mills, but also effective in guaranteeing the stable operation of the system, significantly reducing the fault rate, and improving the production efficiency.