APU feature integration based on multi-variant flight data analysis

Abstract

For aircraft complex systems such as auxiliary power unit (APU), the performance evaluation is currently restricted to observation of several typical parameters. Many other monitoring parameters recorded in Quick Access Recorder (QAR) reflect the APU condition from various aspects yet without enough attention. This study intends to propose integrated performance indicators through feature extraction among many monitoring parameters. Clustering analysis is then conducted to validate the effectiveness of the method by anomaly identification. This method has the potential to easily evaluate performance of some complex aircraft systems for early warning and prevent degradation from early stage.