Research On Early Fault Detection Of Hot Components In Gas Turbines

With the development of gas turbine technology,the combustor outlet temperature of gas turbine is getting higher and higher.Flame cylinder,transition section,first stage turbine nozzle,moving blade and other hot components not only bear extremely high temperature,but also need high fracture strength,thermal mechanical fatigue,creep resistance,oxidation resistance,corrosion resistance and other properties.Its manufacture and repair technology is extremely difficult.It is the most critical component of a gas turbine.Once the hot components are damaged,great economic losses will be caused.Therefore,it is urgent to study the early fault detection method of gas turbine hot components,find an early fault indicator of hot components in advance,and take preventive measures to avoid the occurrence of hot components damage accidents.In this paper,the key problems of early fault detection for hot components of gas turbines are analyzed,and the early fault detection method for hot components of gas turbines are studied.First,the key problems of early fault detection of gas turbine hot components are analyzed.It is found that the difficulty of early fault detection for hot components of gas turbine is that besides the fault,the normal operating conditions of gas turbine,such as noise,operating conditions,inherent structural deviation,gas rotation and mixing,have an impact on the exhaust temperature distribution,resulting in excessive in-class distance of normal exhaust temperature distribution.The solution of early fault detection for hot components is proposed,that is,to eliminate the influence of the above factors on exhaust gas temperature distribution,and to compress the intraclass distance to improve the detection performance of early fault detection.Second,a gas turbine model with multiple combustors is established,and the mixing and rotating effects of hot gas in the turbine are considered.The model can reflect the typical hot component faults,the inherent structural deviation of hot components,and the influence of noise,operating conditions and gas rotation on exhaust gas temperature distribution,and meet the needs of early fault detection on hot components.The model lays a foundation for the follow-up research on early fault detection of hot components.Third,the relative exhaust gas temperature deviation index is extracted,which reflects the inherent structural characteristics of hot components of gas turbine and does not change with operating and ambient conditions.An early fault detection method based on relative exhaust gas temperature deviation index for hot components is proposed,and the validity of the method is verified by simulation data and actual data.The characteristics of noise and the effect on early fault detection of hot component failure are analyzed.Fourth,two early fault detection methods for hot components are proposed,which further eliminates the influence of swirl effect of exhaust gas temperature distribution on the detection.Based on the quantum particle swarm optimization(QPSO)method,the continuous exhaust gas temperature distribution along the circumferential direction of the turbine outlet and the continuous relative exhaust gas temperature deviation index along the circumferential direction of the turbine outlet are obtained by introducing the prior knowledge of the gas turbine.The swirl effect of the exhaust gas temperature distribution is considered.At the same time,an early fault detection method based on convolutional neural network(CNN)for hot components of gas turbines is proposed.The advantages of CNN in solving the problem of exhaust gas temperature distribution swirl are analyzed in detail in principle.The detection method is visualized.Fifth,the complete early fault detection method of hot component is proposed and is applied in practice.The results show that the proposed method has strong sensitivity and robustness,and can effectively eliminate the influence of operation and ambient condition changes,gas swirl and other factors.It can detect the fault of gas turbine hot components as soon as possible,and realize the early fault detection of gas turbine hot components.