Ziqiang Pu

Zhengzhou University of Light Industry, University of Algarve

Papers

Total Citations

H-Index

About

Ziqiang Pu is a researcher whose work sits at the intersection of machine learning, generative modeling, and intelligent fault diagnosis systems. His research addresses one of the most persistent challenges in real-world industrial applications: the problem of imbalanced data, where normal operational samples vastly outnumber anomaly instances, making accurate fault detection exceptionally difficult. Pu's most impactful contribution is his development of generative adversarial network (GAN)-based frameworks tailored specifically for fault diagnosis. His 2021 paper introducing a One-Class Generative Adversarial Detection Framework has garnered 55 citations, demonstrating significant community recognition for its multifunctional approach to diagnosing faults under unbalanced data conditions. Building on this foundation, his 2022 work on VGAN — a principled generalization of both MSE GAN and WGAN-GP architectures — addresses critical training instabilities such as mode collapse and oscillating convergence, with direct applications to robot fault diagnosis, accumulating 23 citations. Together, these contributions reflect Pu's sustained effort to make fault diagnosis systems more robust, generalizable, and practically deployable. His focus on bridging theoretical GAN advancements with engineering applications positions him as a meaningful contributor to the growing field of AI-driven system health monitoring and predictive maintenance.