Research On Fault Diagnosis Ofaero-engine Rotor System Based On WaveCluster

With the rapid development of the aviation industry, people have paid more andmore attention to the operational reliability of aircraft. The minor aircraft accident maycause normal flight of the aircraft. And, if the aircraft accident is serious, then it maycause aircraft damage or even personnel sacrifice. Whether it is the above situation, itwill result in significant losses. Aero-engine is the most important part of the aircraftand has been hailed as the "heart" of the aircraft. It is particularly important fortrouble-free operation of the core components of the aero-engine rotor system.According to incomplete statistics, about40%of the flight accident was caused byengine failure, while30%of the mechanical failure was caused by the failure of therotor system. Therefore, the fault diagnosis methods for aero-engine rotor system haveimportant social and economic benefit.The wavelet-based clustering method (wavecluster algorithm for short) is totallyinsensitive to the ordering of the input data objects and the clustering result can freefrom noises. It can discover clusters of any arbitrary shape on different scales and doesnot need users to decide the parameter about the number of clusters. In addition, itsrunning speed is very fast. Therefore, wavecluster algorithm is focused and the researchon fault diagnosis of aero-engine rotor system based on wavecluster is proposed in thispaper.In this paper, firstly, the research situation of clustering method is introduced, thecommonly used clustering algorithm performances are compared and the reason isgiven about choosing the wavecluster algorithm. Then, the domestic and overseasresearch situation of wavecluster algorithm is simply introduced. Secondly, on the basisof the principle and type of aero-engine, the common faults of aero-engine are leaded.Then, fault reasons and vibration characteristic of four common aero-engine rotorfailures are introduced in detail. Thirdly, the wavecluster algorithm is detailedlyintroduced, and its difficulties are analyzed. Finally, the rotor faults data are used toverify the feasibility and effectiveness about wavecluster in the study of aero-enginerotor system fault diagnosis. In the quantize feature space process, the method ofcombining with empirical formula and heuristic is used to solved the MAUP problem. The calculation results show that this method can quickly determine the optimalquantize values. In the process of finding the connected components, the4-connectiondefinition is extended from the2dimensional feature space to the d dimensional featurespace, and it is successfully applied to the8dimensional feature space. The finalcalculation result shows that the fault diagnosis method of aero-engine rotor systembased on wavecluster is very effective, and the rotor faults data can be assigned to theappropriate cluster.