Method of aero-engine flight safety pre-warning based on support vector machine

Abstract

Abstract Engine vibration is a decisive factor in flight safety. Aiming at the problem that the current single vibration warning value of an aero-engine cannot reflect the vibration characteristics of the engine in flight, the estimated probability density of the measured vibration of the engine based on the support vector machine (SVM) method is established, and the aero-engine flight safety pre-warning research is carried out. Through the statistical analysis of the installed flight data of a engine, the probability density function estimation of the measured vibration data of the aero-engine flight was carried out by the support vector machine, and the vibration warning value of the installed engine in each state was determined according to the probability distribution and the general primary side determined by the vibration monitoring limit value of rotating machinery. Through the verification of actual flight data, the determined vibration warning value with the change trend of engine state is consistent with the actual flight data, and is more sensitive to engine failure, which is conducive to improving the early failure warning ability of aero-engine during flight.