MultiCogniGraph: A multimodal data fusion and graph convolutional network-based multi-hop reasoning method for large equipment fault diagnosis

Abstract

As industrial production escalates in scale and complexity, the rapid localization and diagnosis of equipment failures have become a core technical challenge. In response to the demand for intelligent fault diagnosis in large-scale industrial equipment, this study presents “MultiCogniGraph”—a multi-hop reasoning diagnostic method that integrates multimodal data fusion, knowledge graphs, and graph convolutional networks (GCN). This method leverages internet of things (IoT) sensor data, small-sample imagery, and expert knowledge to comprehensively characterize the equipment state and accurately detect subtle distinctions in fault patterns. Utilizing a knowledge graph to synthesize data from multiple sources and deep reasoning with GCN, “MultiCogniGraph” achieves swift and effective fault localization and diagnosis. The integration of these techniques not only enhances the efficiency and accuracy of fault diagnosis but also its interpretability, marking a new direction in the field of intelligent fault diagnostics.