Rolling bearing fault diagnosis with variable load and few samples based on multifeature fusion meta-learning

Faced with variable working conditions and limited samples, the fault diagnosis of rotating machinery rolling bearings in the actual industrial process will result in reduced accuracy of intelligent fault diagnosis methods based on deep learning. In this paper, to solve the existing problems faced by bearing fault diagnosis, a fault diagnosis model based on multi-feature fusion Model-agnostic meta-learning is proposed in a few-shot variable load scenario, and the bearing multi-modal feature fusion and channel attention mechanism are introduced. Through domain adaptation experiments on bearing data set provided by Case Western Reserve University, the effectiveness and accuracy of the method in bearing fault classification are verified.