ABSTRACT
Intelligent fault diagnosis is essential for downtime reduction in modern industries. However, domain (i.e., working condition) variants are unavoidable for fault diagnosis under changing environment. Differentiable architecture search (DARTS), as an automatic machine learning technique, can automate the network design process for a specific domain efficiently by using hyper-network and differentiable search strategy. Nevertheless, the representation of hyper-network is restricted by several unfair factors, and the searched architecture of DARTS is overconfident. To address these issues, a one-shot neural architecture search approach, which involves two-stage learning, is proposed for efficient domain matching fault diagnosis. In the first training stage, the warmup and path-dropout strategies are taken to enhance competitiveness of the parametric operators and alleviate the co-adaptation problem to obtain an intuitively fair hyper-network. In the second matching stage, variational inference is introduced into differentiable search strategy to estimate the uncertainty of model matching, and a scale mixture prior is used to softly constrain the matching stage. Multi-domain experiment is implemented by adding noise to the raw signal, and the proposed method outperforms four commonly used deep neural networks for aeroengine bevel gear fault diagnosis.
