Research On Fault Diagnosis Of Gas Turbine Rotor System

With the implementation of the West-East Gas Pipeline Strategy and the continuous utilization of natural gas resources,as an efficient and clean powered device,gas turbine is widely used in industrial production.Rotor is the most important part of gas turbine,so its reliability and safety become more and more important,which poses a higher requirement for the accurate and rapid diagnosis of faults in gas turbine rotors.In a word,it is very important to analyze the vibration signal of the rotor and judge whether the operating state of the gas turbine is normal or not for protecting safety and increasing productivity.In this Thesis,the common faults types,the mechanisms of faults and the corresponding characteristics of gas turbine in the running state are analyzed,as well as the vibration characteristics of the gas turbine rotor in different faults are described.Based on the study of the related principles and methods of rotor fault diagnosis,the vibration signal analysis method based on Fourier Transform is used to judge the running state of the gas turbine rotor.The vibration monitoring system of Rockwell Automation’s Entek is applied for the fault diagnosis of the gas turbine rotor of a compressor station,and the experimental results show that the Fourier Transform has some limitations in the analysis of nonlinear vibration signals.To improve the accuracy of gas turbine rotor fault diagnosis,this thesis proposes Support Vector Machine(SVM)for gas turbine rotor fault diagnosis.Because of the large number of feature variables involved in decision-making,principal component analysis(PCA)is used to extract the main feature variables before the fault diagnosis of the gas turbine rotor.Considering that the selection of parameters of SVM has a great influence on the results of fault diagnosis,the PSO algorithm is adopted to optimize the kernel parameters.At last,rotor fault simulated experimental box is used to extract the typical fault data of the rotor,and the support vector machine(SVM)is used to diagnose the fault.The experimental results show that the fault diagnosis method of gas turbine rotor based on SVM can distinguish fault types,which achieves the purpose of improving the diagnosis effect of rotor failure.