Beyond single prototype: A multi-prototype fusion framework with contrastive learning and diffusion model for zero-shot fault diagnosis

doi:10.1016/j.inffus.2026.104399

Beyond single prototype: A multi-prototype fusion framework with contrastive learning and diffusion model for zero-shot fault diagnosis

2026-04-22

0

PRE

AI

DOI：10.1016/j.inffus.2026.104399

Original

Original request for help

Save

Abstract

En 中文

• A prototype fusion module is developed to solve the zero-shot domain shift problem. • A fusion contrastive learning is proposed to enhance feature separability. • A new diffusion model is proposed to enhance the stability of generated samples. • The proposed model can identify compound faults only with single fault samples.

Keywords:

multi-prototype fusion

contrastive learning

diffusion model

zero-shot fault diagnosis

compound fault identification

Beyond single prototype: A multi-prototype fusion framework with contrastive learning and diffusion model for zero-shot fault diagnosis

Abstract

Journal

Organization