An innovative approach based on meta-learning for real-time modal fault diagnosis with small sample learning

The actual multimodal process data usually exhibit non-linear time correlation and non-Gaussian distribution accompanied by new modes. Existing fault diagnosis methods have difficulty adapting to the complex nature of modalities are unable train models based on small samples. Therefore, this paper proposes a modal method meta-learning (ML) neural architecture search (NAS), MetaNAS. Specifically, best performing network model existing is first automatically obtained using NAS, then, design learned from NAS ML. Finally, when generating modalities, gradient updated experience, i.e., quickly generated under sample conditions. effectiveness feasibility proposed fully verified numerical system simulation experiments Tennessee Eastman (TE) chemical process. As primary goal, abstract should render general significance conceptual advance work clearly accessible broad readership. References not be cited in abstract. Leave Abstract empty if your article does require one–please see “Article types” every Frontiers journal page for full details.