Analysis Method Of Railway Vehicle Axle Acoustic Emission Data Flow Characteristic

Mechanical failure is common in real life.And to ensure the safety of the production and normal production activities,all kinds of fault detection method arises at the historic moment.In the fault detection,there are two kinds of methods which are destructive testing and nondestructive testing.The acoustic emission signals detection method is a nondestructive testing method which cannot only protect the testing machine,and as a kind of endogenous signal,acoustic emission waves generated by the object itself,which can avoid additional external detection signal source like other nondestructive testing methods,such as ultrasonic.So,the acoustic emission phenomenon and the corresponding detection method has received the widespread attention.This paper studies how to use the acoustic emission data flow to analyzethe whole process of railway vehicle axles from crack initiation,crack extension to final fracture.First of all,the use of acoustic emission phenomenon for failure analysis needs to distinguish whether the collected signal belongs to the acoustic emission signal.This paper proposes two methods of acoustic emission signal recognition,one kind is suitable for online fast acquisition and analysis by analyzing time domain features of acoustic emission signals,another kind is suitable for offline accurate analysis by analyzing three-dimensional energy featuresof acoustic emission signals.In the acoustic emission signal online time-domain analysis method,this article introduces the basic concepts of data flow andbasic knowledge of time domain features.Then based on the traditional time domain signal analysis method,combined with the conceptof data streams,this paper puts forwardonline acoustic emission data stream time domain featurescalculation method and neural network recognition method with a feature waiting pool,which is called STREAM-TIME-ENN model.Data flow methodimproves the computing precision with no obvious increase calculation,and feature waiting pool method significantly reduces the time consumption of identification process.And then the experiment shows the value of the algorithm in the actual environment.In the acoustic emission signal energy domain analysis method,this paper introduces the empirical mode decomposition method and Hilbert transform method,and the research status of acoustic emission signal energy domain analysis.Then,on the basis of the traditional analysis method,this paper summarizesathree-dimensional energy feature analysis system,whichis called EMD-TDE-ENN model and describes the acoustic emission signal energy domain features comprehensively.By comparing with the traditional methods,this method is better in describing the acoustic emission signals.