Research On Undetermined Blind Source Separation Based Aircraft Engine Fault Location And Diagnose

Aircraft engine is the heart of the aircraft.Its safety is a powerful guarantee for a reliable and safe flight.Therefore it's significant to monitor the state of the engine and diagnose the fault in the first place.In this paper,the undetermined blind source separation technique and the multiple model estimation fault diagnose method are studied to realize the fault location and diagnosis,thus to provide support for the operation and maintenance of the aircraft engine.Followings are the main work of the paper.(1)Considering the fact that the number of the sources is often more than that of measured points,a method of using the undetermined blind source separation algorithm to separate the measured signals is proposed.By optimization and improvement for the classical undetermined blind source separation,algorithms of dual and three channels are presented,which are verified by simulation signals,signals from aircraft engine dynamic model and signals from the operation experiment of an aero turbo shaft engine.(2)In this part,multiple model estimation is applied in aeroengine fault diagnosis.Based on the kinetic equation of common faults in aeroengine rotor,the multiple model estimator consists of a group of Calman filters,who generate conditional probability from the estimated value and the observed signal,thus to identify the state of the rotor and diagnose the fault.(3)Combined with the undetermined blind source separation and multiple model estimation,the signals from engine dynamic model are separated into two main vibration sources,high and low pressure rotor,which are input to the multiple model estimator constructed.Through probability estimation,the preliminary fault location and diagnosis is realized.