Study On The Damage Analysis And Fault Identification For Mechanical Structures

Damages in mechanical structures will shorten the life cycle of the machine andreduced it reliability. The damage under the continuously fatigue load furtherexpands into huge potential safety hazard. Therefore, the analysis of the mechanicalstructure damage, identification of the fault, intervention of the equipment andtroubleshooting have a very important significance.The crack generated on the surface of the mechanical structures under badcondition often results in an accident, considering the aero-engine rotor blade isexpensive and inconvenient in removing, the damage sheet under loading hasbecome an important object in our research. Faults happening on aero-engine rotorwill cause the resonance of the pipe, the pipe will be in the abnormal state. Forvibration signals in different fault types,its entropy is also different. And some faultsare generally at a particular frequency, so we need to diagnose the faults, identify thetype of the fault, and further take interventions to the equipment in order totroubleshooting. Thus damage detection is of great practical significance inmanufacturing and engineering applications.The main contents in this paper include the following three points:1) Spectrum Analysis of strain signals on mechanical structures Based on FBGand Hilbert Transform. The signal acquisition of dynamic strain in the rotating bladesof aircraft engine is introduced, then a spectrum analysis on straining signals ofrotating blades has been done through fiber Bragg grating sensors (FBGs).As a result,we extract the features in spectrum, in allusion to the frequency components of#3, alocal zero method and HT have been used to assume the cause of#3frequencyclusters. According to the measured and analyzed results, the FBGs are effective tomeasure the dynamic strain of blade in operating aero-engine. Moreover, it canprovide the state of rotating blades based on the envelope analysis in HT.2) Damage detection based on multi-scale entropy and support vector machine.A damage detection method based on multi-scale entropy (MSE) and support vectormachine (SVM) is presented to analyze the damages on a vibrated mechanicalstructures in our experimental platform. According to the measurement and analyzed results, this method is effective to determine the minimum area containing thedamages on the basis of points detected by the FBGs and it is better and moreaccurate than the SFRF index, however, the type of damage is an important factor thataffects different area on the plate.3) A method based on BBS and MSE is proposed to monitor the operating stateof the mechanical structures, firstly, the experiments of aero-engine pipelines isintroduced, and then JADE is used for the normal signals of aero-engine pipeline toreduce interference,then, the signal acquisition of the aero-engine pipeline isintroduced, next, four kinds of simulated fault signals of rotor are added to normalsignal of aero-engine pipelines and the effectiveness of the method can be verified bycontrasting the area of MSE of each signals including the normal pipelines signal.Then, in order to verify the effectiveness of the method, we change the size of theinput voltage and the proportion of the amplitude of the rotor fault signal and thereference signal, by observing the different scales of MSE values of the four kinds offault signals, we can distinguish four kinds of faults.