Performance Comparison of Blind Separation Algorithms for Aero-engine Vibration Signals

摘要

When the machine is operating, the sensor signals are mixed by every vibration source, which cause great difficulties to machine health monitoring and fault diagnosis. Blind source separation (BSS) technique can separate the mixed signal without losing the characteristic information of source signals. To find the best algorithm for separating the practical vibration signals, five typical algorithms have been studied and compared. Five typical algorithms were used to separate the simulation signals of mechanical vibration, the performance of five algorithms under different environments were compared using different performance index for evaluating the separation results. BSS was used to separate the mixed rotor's vibration signals, the result shows that JADE algorithm can separate the mixed vibration signals successfully.