Identification of key factors affecting the failure of aviation piston engine turbochargers based on an improved correspondence analysis-polar angle-based classification

Abstract

Turbocharging is an efficient approach for addressing power reduction and oil consumption increase in aviation piston engines during high-altitude flights. However, a turbocharger significantly increases the complexity of a power system, and its considerably complex matching relation with the engine results in a coupling of failure modes. Conventional analytical methods are hard to identify failure-inducing factors. Consequently, safety issues are becoming increasingly prominent. This study focuses on methods for identifying failure-inducing factors. A whole-machine system model is established and validated through experimentation. The response surface method is employed to further abstract the system simulation model to a surrogate model (average error: ~ 3%) in order to reduce the computational cost while ensuring accuracy. On this basis, an improved Correspondence Analysis (CA)-Polar Angle (PA)-based Classification (PAC) is proposed to identify the key factors affecting the failure mode of turbochargers. This identification method is based on the row profile coordinates G varying with the numerical deviations of the key factors, and is capable of effectively identifying the key factors affecting the failure. In a validation example, this method identifies the diameter of the exhaust valve (e2) as the primary factor affecting the safety margin for each work boundary.