Fault detection of a reusable rocket engine using phase plane trajectory of feature vectors

Abstract

A bivariate time-series analysis based on the phase plane trajectory of feature vectors extracted by principal component analysis is developed for fault detection in a reusable liquid-propellant rocket engine. Static-firing test results of the reusable rocket engine obtained at the Japan Aerospace Exploration Agency are employed for demonstration of the present method. The present method successfully detected temperature sensor failure from 19 firing tests of 62 sensors, even in the deviation of the engine operational sequence between the static-firing tests. The present method was also able to detect the system failure from 23 firing tests. Furthermore, the ability to distinguish the system and sensor failure was demonstrated.