The Vibration Fault Analysis Of Aero Engine High Pressure Rotor

Aero-engine is the heart of airplane, and its structure becomes more and more complex,the thrust, rotor speed and intensity of work is improving, which lead to aero-engine vibration faults significantly increased. There are many factors causing aero-engine vibration fault, but high-pressure rotor vibration fault is one of the most important reasons. Therefore, the vibration failure analysis of aero-engine high pressure rotor becomes more and more important.The analysis process of high pressure rotor vibration fault includes three parts: the acquisition of vibration fault signals, the extraction of vibration fault characteristics and the identification of vibration fault. Among them, the extraction of vibration fault characteristics is the key. In this paper, a new method of the time-frequency analysis, the empirical mode decomposition(Empirical Mode Decomposition, abbreviated EMD) and the new pattern recognition technology, support vector machine(Support Vector Machine, called SVM), are applied for vibration fault diagnosis of the high pressure rotor.We summarize the vibration characteristics of four common aero-engine vibration failures, establish a mathematical simulation model of vibration fault signals, describe the signal analysis technology of aero-engine high pressure rotor vibration faults, and systematically study two aero-engine high pressure rotor vibration faults diagnosis methods:Mahathir Laurin Biscay distance function fault diagnosis method, through analyzing the vibration fault signals of aero-engine high pressure rotor by applying the empirical mode decomposition, extracting their energy eigenvalues of Hilbert marginal spectrum, using Mahathir Laurin Biscay distance function to calculate the distance between the vibration fault signal and the sample signal, according to the minimum distance principle, determine the type of vibration fault; support vector machine fault diagnosis method, establishing the auto regression models of high pressure rotor vibration fault signals, using the least squares method to determine the model parameters and residuals, constituting a feature vector by using the model parameters and residuals, importing them to support vector machine multi-class fault classifier, and determining the vibration fault type based on the output.