Use of artificial intelligence to classify failure data from a military aircraft fleet

Purpose To develop and evaluate an artificial intelligence (AI) model capable of automatically performing failure classification for Brazilian Air Force aircraft repair data. Timely, accurate failure data is crucial for effective lifecycle management in support of mission readiness, enabling improved resource allocation and aircraft availability, and reduced maintenance costs. Design/methodology/approach Textual data in aircraft repair records were processed using natural language processing (NLP), and six different machine learning techniques were used to developed models to classify these records. The models were evaluated and the best performing model used to classify new data. These classifications were then compared against manual classifications. Findings The selected model, a Support Vector Classifier, classified new data with an F1-score of 95.8%. And in some cases where the model misclassified records, a reexamination of the data revealed that the manual classification was actually in error. Originality/value Very limited literature exists that applies AI to unstructured military aviation maintenance data. This study expands the use of AI in managing and processing aircraft maintenance data by employing NLP and AI to classify repair data across a variety of aircraft types, producing a more reliable and generalizable model. This study also describes implementation of the model by the Brazilian Air Force and plans to establish an AI LAB to manage military data.