Data Mining-Based Analysis of Alert Messages of Executive Aircraft

Once the years, a large number of alert messages have been stored by aircraft manufacturers, which characterize the aircraft behavior and performance. This way, some aspects such as component failures and aircraft health can be obtained by means of these messages in order to make decisions and better manage the aircraft. However, these messages are stored faster than can be processed by the reliability engineering team. Motivated by this fact, this paper proposes the use of Data Mining to determine sequential patterns of messages. In this context, it can be observed that the use of Data Mining may contribute to prevent aircraft accidents and reduce the time of maintenance. Therefore, this paper presents a methodology based on the Apriori algorithm to extract association rules from a database of messages sent by executive aircraft operating in Latin America. It is important to highlight that the proposed method was validated using a database provided by a real aircraft manufacturer. Finally, the results obtained by the Apriori algorithm were analyzed in contrast with specialist's knowledge, for known messages, where the method robustness could be validated since the associated messages always had a systemic cause for such emergence.