Multi-sensor based graph convolution fault diagnosis method

In recent years, research based on Graph Neural Network (GNN) has made significant progress in the field of rotating machinery fault diagnosis. Current methods usually only rely on data from a single sensor for diagnosis. This method is easily affected by problems such as noise interference and missing data from the sensor itself, resulting in inaccurate diagnostic results. Multi-sensor systems can reflect the operating status of machinery more comprehensively and reliably, and also put forward higher requirements for data algorithms. Different from previous deep learning models, graph neural networks have significant performance in mining graph structures and can effectively integrate multiple node relationships and features. Therefore, this paper proposes a multi-sensor based graph convolution fault diagnosis method. This method first converts data collected by multiple sensors into a graph data structure, and then uses a graph convolutional neural network to process these graph data and extract feature information. Finally, the feature information is input into the classifier for fault diagnosis. Compared with the single-sensor fault diagnosis method, this method has higher accuracy and reliability.