Research On Gas Path And Sound Diagnosis Method For Gas Turbine Health Status

As the gas turbines play an increasingly important role in the aerospace,marine,and power industries,their safe and efficient operation is increasingly important to system safety and economics.Due to the complexity of the gas turbine structure and the harsh working conditions,the gas turbine cannot be operated in a healthy status.Therefore,in order to improve the reliability of gas turbine operation and reduce accidents,it is necessary to conduct a health status diagnosis study on gas turbines.In this paper,the gas path diagnosis method and acoustic diagnosis method are used to carry out research on the health status diagnosis of gas turbines.The main work includes:The objects of gas turbine health status diagnosis research: performance degradation status of compressor and turbine,bearing vibration status of gas turbine rotor system,and combustion combustion oscillation status of gas turbine.In this paper,different diagnostic methods are needed for the health status diagnosis of different objects of gas turbine: using gas path diagnosis method to analyze and diagnose the degradation status of gas turbine compressor and turbine gas path,and using acoustic diagnosis method to analyze and diagnose the vibration status of bearing of gas turbine rotor system,and using acoustic diagnosis method to diagnoses and analyzes the combustion oscillation status of the combustion chamber of the gas turbine.This paper studies the application of gas path diagnosis method in the diagnosis of gas turbine performance degradation status.Based on the operational sample data of the target gas turbine,a target gas turbine performance degradation model based on thermodynamics method,a target gas turbine performance degradation model based on BP neural network,a target gas turbine performance degradation model based on a method combined the BP neural network and thermodynamic method are established.The gas turbine performance degradation model based on thermodynamic method is used to analyze the effect of gas turbine compressor and turbine performance degradation on gas turbine performance parameters.The gas turbine performance degradation model based on BP neural network and the gas turbine performance degradation model based on a method combined the BP neural network and thermodynamic method are compared.The verification calculation results of the simulation model of gas turbine performance degradation have the following conclusions:(1)For the overall gas turbine performance,the degradation in flow is more sensitive than the degradation of efficiency,and the degradation of turbine performance is more sensitive than the degradation of the compressor.(2)The calculation accuracy of the gas turbine performance degradation model established by the method combined the BP neural network and thermodynamic method and the gas turbine performance degradation model established by BP neural network can meet the requirements of gas turbine performance degradation simulation calculation.But the former has a higher accuracy,the calculation error is on average 1% lower than the latter.And the latter has the advantage of more convenient calculations,only need to establish a BP neural network.(3)The gas turbine performance degradation model based on thermodynamics method can be used to diagnose and analyze the effect of compressor and turbine performance degradation on the performance degradation status of the gas turbine,but its calculation accuracy is lower than the performance degradation model of the gas turbine established by BP neural network.The application of acoustic diagnosis method in the diagnosis and analysis of bearing vibration status of gas turbine rotor system is studied,and the bearing vibration status test bench is designed and established.During the test,the acceleration signal and sound signal on the bearing in vibration status due to the wear of the bearing components(inner ring wear,outer ring wear and rolling element wear)and the bearing in the normal status were collected.And the acceleration signal is used to verify the sound signal in the bearing vibration status diagnosis.From the test results,it can be concluded that:(1)The time-domain characteristics of sound signal and acceleration signal can basically distinguish whether the bearing vibration occurs or not,but the specific vibration causes of the bearing can not be distinguished,so it is suitable for diagnosing whether the bearing vibration occurs in operation.(2)The time domain and frequency domain characteristics of bearing vibration signals can be obtained by wavelet packet decomposition and reconstruction.(3)After the decomposition of the sound signal and the acceleration signal in the wavelet packet,although the energy spectrum distribution is inconsistent,the maximum sub-band of the difference between the vibration state and the normal state energy spectrum is consistent,and the main frequency characteristics of the reconstructed signal are consistent from the reconstruction of the sub-band.In the wear of outer ring and rolling element,the acceleration signal will be affected by the interference signal in the area higher than 600 Hz,while the sound signal will be less disturbed,that is,the anti-interference ability of the sound signal is strong.Therefore,the acoustic diagnosis method can be applied to the diagnosis and analysis of bearing vibration status,and compared with the acceleration signal,the acoustic diagnosis method has the advantages of non-contact,anti-interference and so on.This paper studies the application of acoustic diagnosis method in the diagnosis of combustion oscillation status of combustion chamber.The test benches of LPP combustor and DLN combustor were designed and built.The dynamic pressure signals and sound signals were collected and analyzed by combustion oscillation test.Through the comparative analysis of the characteristics of sound signal and pressure signal,the feasibility of applying acoustic diagnosis method to the diagnosis of combustion oscillation status in combustor and its advantages and disadvantages are studied.The following conclusions can be drawn from the analysis:(1)In LPP combustor,combustion-induced combustion oscillation is selfexcited oscillation,so the combustion status of combustor can be diagnosed and analyzed by analyzing the sound and pressure signals of combustor.Under the condition that the inlet air flow rate is constant,as the fuel flow rate increases,the sound and pressure of the combustion chamber have the same trend of the main frequency value and the amplitude change.Under the condition that the fuel flow rate is constant,as the air flow rate increases,the sound and pressure of the combustion chamber have the same trend of the main frequency value and the amplitude change.The acoustic diagnosis method can be used to diagnose the combustion oscillation of LPP combustor.(2)In the LPP combustion chamber,the sound signal is more sensitive than the pressure signal to the combustion process of the combustion chamber,and can be used to analyze and predict combustion oscillations.The pressure signal can only reflect the main frequency change of the combustion oscillation in the combustion chamber,and cannot reflect the ignition and flameout combustion process.(3)In the DLN combustion chamber,the preheating condition of the combustion inlet air can reduce the oscillation amplitude of the combustion oscillation and increase the main frequency value of the combustion oscillation,that is,the oscillation intensity in the combustion chamber can be reduced.(4)In the DLN combustion chamber,the characteristics of the sound signal and the pressure signal are consistent with the change of the equivalence ratio in the combustion chamber.The amplitude of the sound signal between combustion oscillation and combustion stability differs by 30%,while the pressure signal is only 10%,indicating that the sound signal is more sensitive to changes in operating conditions than the pressure signal.At the same time,the sound signal reacts to the main frequency value of the combustion oscillation in the combustion chamber earlier than the pressure signal,that is,the sound signal is more sensitive than the pressure signal,and the main frequency signal of the combustion oscillation can be analyzed earlier by sound signal.Through the above work,the application of gas path diagnosis method and sound diagnosis method in gas turbine health status diagnosis is studied in this paper.The BP neural network and thermodynamic method are combined to calculate the gas turbine performance degradation model with higher accuracy and acoustic diagnosis method,which is sensitive,easy to collect and strong in anti-interference are obtained.