Information Extraction Of Deformation Based On Wavelet Packet Transform

Deformation is one kind of natural phenomenon, which means thechanges of shape, size and location of deformable bodies, caused by variousfactors. Once the value of deformation exceeds its allowable value, somedamage will be caused. As the deformation is changing all the time, it isnecessaryto observe the deformable bodies so as to know about the situationofdeformation.However as aresult oftheinfluenceofsubjectivefactors andobjective factors, the data obtained from observation contains all kinds ofnoise. In order to obtain the real information of deformation, we needeffectivemethodstoeliminatethenoises.Wavelet packet analysis theory, which is the extension of waveletanalysis theory, is also a signal processing method. It is developed rapidly inrecent years. It can analysis signal more particularly than wavelet method. Itcananalyzesignal inall frequencydomain,andit is applied broadlytosignaland image processing. In this paper, deformation data is regarded as a signal.Wavelet packet analysis method is applied to the deformation signal. Theworkisdoneasfollows:(1) The main techniques for obtaining deformation data is introduced,and the influence factors is indicated.Then the methods in deformation dataprocessing are introduced, and the advantage and disadvantage of them isdiscussed.(2) Wavelet theory and wavelet packet theory are introduced. And applicationinsignalprocessingisdiscussed.(3) The application of wavelet package theory in de-noising fordeformation signal is discussed. The basic de-noising theory and de-noisingmodel based on wavelet package is introduced. Furthermore, based onoriginal threshold de-noising model, an advanced threshold de-noisingmethod is proposed. The feasibility of the new threshold is verified by usingsimulatingandrealsignal,andobtainsexcellenceresult.(4) The application of wavelet package theoryin extracting deformationinformation is discussed. For the characteristic, that the deformation data isinfluenced by various disturbing factor. Spectrum analysis method is used togainthemain frequencyfeatureinthesignal.Thenthedataisdecomposedinall frequency domains by wavelet package method, and choose the properwavelet package basis in that the main frequency band lies. Finally, thedeformation information is extracted by feature analysis for time andamplitude of signal. The information of deformation extracted by uppermethodistestedtobegood.