Analysis and Neuro-Fuzzy Algorithm

Abstract

Pipeline .si icfke defects such ns csncks cnii.se rnqjor prob1enr.v ,fils mssei inanagers, pnsliculns!v wi7eii ilie pipe is b u r i d under ihe ground. Tlie innnunl rn.speciion (f surface (I. f2ci.s in the underground pipes has (1 ni i i i iher of drawhncks;.s, iiicliudiiig subjectivitv, vniyviiig .statitlrrtls, N I I ~ high cos~s. Aii~oirtnric inspectiort , processing and r?rr(fjcia/ intel l tgn?~~~ lecliniqiics can overcoim mari~v of these disadvnntnps and o/]i:r oxset managers an opporiunity to ,signi ficonIlv it?ipsow qunli!y and reduce costs. .4 recognition nnrl clnssificniion of' pipe cracks using ininge nnalysis and tieusoTfuzzv nlgoriihiri is proposed 61 t i le pre-processing siep. the crnc:k.\in the pipe are extsnctecl ,jPom the hoiiro~geiious hnck,.round Then, hnsetl on N prior knowler[yc of crnckx, ,jive normalised , featuscs are extracted 117 the cla.v.v~Jicniion step. n neuro-fiizq: algorithni is proposed that eniplqvs a trapezoiclal, fuzTv nienibership functiori nnd irrocli f i cd ersor backpropagnlioii (El%') algorithnr.

6 Figures and Tables

Cite this paper

@inproceedings{Sinha2004AnalysisAN, title={Analysis and Neuro-Fuzzy Algorithm}, author={Sunil K. Sinha and Fakri Karray and Paul W. Fieguth}, year={2004} }