Information Extraction of Aircraft Maintenance Records for Knowledge Graph Construction

Maintenance-workforce efficiency has a profound effect on cost. Inefficient maintenance operations will spend more on maintenance labor to complete the same amount of work. Maintenance records involve work history, including critical information on failure modes, which can be used to guide maintenance engineers in aircraft maintenance processes. For improving the utilization efficiency of maintenance experience, this paper proposes a joint learning cascade binary tagging framework to extract necessary knowledge in the unstructured maintenance record provided by China Eastern Airlines. A knowledge graph for fault diagnosis of civil aircraft environmental control system is constructed after obtaining knowledge. The method is beneficial for improving the efficiency of fault location, optimizing maintenance operations, and reducing costs for airlines.