Hybrid Modeling Of Gas Turbine And Application In Gas Path Fault Diagnosis

Gas path faults of gas turbines can change the performance of the components,which belong to the frequent fault type of a gas turbines,and have a great impact on its safety,reliability and economy.At present,model-based performance analysis is the main method for gas path fault diagnosis of a gas turbine.By analyzing the deviation between predicted data of model and actual parameters of gas turbine,the faults could be judged.The gas turbine model has a significant on the diagnostic accuracy Therefore,It is important to investigate an accurate gas turbine model for its fault diagnosis.However,in a physical application,the accurate characteristic curve of gas turbine is difficult to obtain,and simplification and hypothesis exist in mechanism modeling process.These reasons may lead to large errors in gas engine model,which may affect the accuracy of gas path fault diagnosis.In order to solve the practical problems mentioned above,the idea of applying hybrid model to gas turbine fault diagnosis is put forward in this paper.Based on the mechanism model and data-driven model of the gas turbine,this paper establishes the hybrid models from the system level and the component level respectively,and builds three hybrid model architectures based on series correction or parallel compensation of the whole or the components of a two-shaft gas turbine engine.By comparing the simulation results of the above hybrid models and mechanism model,it is verified that the hybrid models are more accurate.Among the three hybrid models,the system-level hybrid model of the gas turbine has the highest accuracy,because it avoids the error transferamong the components in gas turbine.Based on the principle of model-based gas path performance analysis,a gas path fault diagnosis algorithm based on different models is established with descent optimization algorithm and fault diagnosis criterion The diagnositic results are compared by using actual fault cases of a two-shaft gas turbine engine,such as inlet dust accumulation,scaling and washing,and blade damage.The analysis results show that the fault diagnosis algorithm with hybrid model is more reliable than the mechanism model.It proves that the hybrid model is helpful to improve the reliability of the gas path fault diagnosis of gas turbines.