Semantic integration of heterogeneous aircraft though-life documentation

Abstract

This thesis investigates the challenges of integrating heterogeneous aircraft maintenance records through the development of a novel ontology and prototype application, addressing the complexity and heterogeneity of the data sources, formats, and semantics inherent in aircraft maintenance documentation. This research focuses on the maintenance records of a Boeing 737-400 available at Cranfield to develop Aircraft Maintenance Records Ontology (AMRO). Ontology provides a structured representation of domain knowledge, facilitating improved data consistency by addressing semantic and syntactic inconsistencies. This study aims to minimise manual processes in handling maintenance records, thereby enhancing efficiency, accuracy, and reducing errors related to data inconsistencies. This study outlines an agile development for ontology-based application (ADOBA) methodology, integrating ontology and application development to ensure continuous feedback and improvement. The developed prototype demonstrates practical application in correcting inconsistencies and enhancing record accuracy using Named Entity Recognition (NER) for component categorisation with the corrected data. The performance of the AMRO model and prototype was evaluated using standard metrics, such as precision, recall, and F-measure, confirming their effectiveness in improving data consistency and operational efficiency. The key contributions of this research include a systematic review of aircraft operations and maintenance knowledge management, highlighting challenges related to data heterogeneity and integration, introduction of the ADOBA methodology, and development of AMRO based on real-world maintenance records. This thesis demonstrates the potential of semantic technologies in improving data consistency and operational efficiency in aircraft maintenance, setting a foundation for future advancements and standardisation efforts in the aviation industry. Future work should focus on aligning the AMRO model with international standards, expanding its coverage to include predictive maintenance and lifecycle management, and integrating advanced Natural Language Processing (NLP) models to enhance data interpretation and automation. This research emphasises the importance of robust knowledge management systems to ensure compliance with aviation regulations and improve the overall efficiency of managing aircraft maintenance records.