Study On Key Techniques Of Fault Analysis And Intelligent Diagnosis For Modern Gas Turbine Aero-Engine

Gas turbine aero-engine which is powerplant of modern weapons and equipments (materiel facility) such as avion and missile is a typical complex mechanical system. Its structure is complex, and its working condition is bad. And many kinds of mechanical faults happen easily for it. Many kinds of faults, such as invalidation of engine's gas circuit parts, vibration of rotating parts, wear of friction pair, are all badly affect the security, reliability and high efficiency of its running, so it is very important and significative to improve and perfect condition monitoring and fault diagnosis technique of engine so as to offer technique support for the development of turbojet weapons and equipments.The study and analysis on the faults of gas turbine aero-engine can be divided into two aspects: First, disassemble fault engine, analyze break and crack, evaluate life-span for faults in order to find fault causes and mechanism, thereby the weakness of design can be found and improved on. Where, the fault mode effect analysis (FMEA) is the important method and means for analyzing engine's faults; second, under condition of engine not being disassembled, just check finite detection parameters of fault engine to localize fault, determine the nature and determine the causes. Because gas turbine some complex system characteristics of aero-engine, such as strong nonlinearity, imbalance and uncertainty, lead to the difficulty of modeling and solution based on traditional classical mathematic theories , so the condition diagnosis problems based on multi-source, isomerous, incomplete complex system with indefinite information become more challenging. Moreover, non-classical mathematic methods, such as neural network, genetic algorithm, fuzzy logic, expert system and rough set theories, offer effective approaches to solve these problems.This paper has commenced the study on some pivotal problems about modern gas turbine engine's fault analysis and intelligent diagnosis. Now the summary of main working contents and innovation points in this paper are as follows:(1)The fault mode effect analysis (FMEA) of missile engine is the important part of reliability engineering in the development of engines. By developing the FMEA research, we can make analysis for all potential fault models of engine and study the causes of these faults, and can also find weakness of reliability in design and put forward the corresponding prevention and solution strategy. Finally, engine reliability can be improved and reliable operation of engine in the flight test can be insured. The paper mainly analyses many kinds of fault models happened possibly in the ground and flight test stages of some missile engine development, the causes of fault and the influence for the engine running and flight test, and lists the FMEA table of this type engine sample development stage.(2)Using the FMEA method of missile engine, we can make FMEA analysis for the potential rupture fault mode of high pressure turbine blade. Based on the engine wreckage inspection after disassemble, the break analysis of high pressure turbine rotor blade, engine performance calculations, control system working analysis and engine life-span analysis etc., we have studied the causes and mechanism of fault, and get the conclusions and causes of some engine's high pressure turbine blade break taking place in flight test.(3)In the gas turbine fault diagnosis, we should make full use of much information, but not just pay attention to a sort of information. From the angle of diagnostics, any diagnosis information is ambiguous and inaccurate. And it is half-baked that any single information is to reflect the behavior of any kind of diagnostic object. We should get multidimensional information about the same object from many aspects, fuse and use them, so that we can carry out the more reliable and accurate monitoring and diagnosis for engine. The paper puts forward a kind of fusion diagnosis method based on integrated neural network, and realizes the engine wear fault fusion diagnosis aiming at the wear fault diagnosis problem of gas turbine engine.(4)The expert system based on rules is the best successful field in artificial intelligence application. In this paper, we apply the diagnosis method based on knowledge rule into wear fault diagnosis of gas turbine engine, and build knowledge base, and design reversed reasoning machine, and study the method of knowledge acquisition automatically based on rough set theory. This method overcomes the problem that it is difficult to obtain knowledge in rule-based expert system effectively. At the same time, according to the change of fault symptom in practical case, we advance a dynamic flexible diagnosis method based on extensible knowledge base.(5) Because case acquisition is easier than rule acquisition, so case-based reasoning (CBR) method is the most effective tool to build expert system. In this paper, we study a gas turbine engine troubleshooting expert system based on CBR. Firstly, we build case base; then combining character type fields matching technique with KNN method, advance case searching model aiming at engine troubleshooting cases; finally, we study the case modification method based on difference driving. (6) Because the import and transfer characteristics of gas turbine engine complex system cannot be usually obtained, and it is difficult to forecast engine's export in future. Then we can only use time series analysis method combining with system analysis, and use"system"to process dynamic data, and then obtain system mathematical model. And using this model to identify system and forecast the future development trend can come true. In this paper, using NN forecast method, we advance the trend forecast method based on structure-adaptive NN, and apply it into the trend forecasts of engine performance and wear.