Identification Of Tank Bottom Corrosion Acoustic Emission Signal Based On Fuzzy Clustering

The identification of corrosion acoustic emission signal of tank bottom is the basis ofimproving detection accuracy. However, under some circumstance, the types of corrosionacoustic emission signal of the tank bottom is unknown, a Gauss kernel clusteringrecognition method was proposed to solve this problem.In this paper, based on theoretical analysis and experimental research, carries out thiswork in two areas: The classification of the acoustic emission signal source andoptimization of identification model.Aiming at the randomness of initialize clustering, a Gustafson-Kessel fuzzyclassification algorithm. For improving the accuracy of classification, a method that usingdiscrimination parameter of intra-class compactness and inter-class resolution wasproposed to increase the difference of various acoustic emission signals, optimizedinitialize clustering number. Then the problems of local optimal and sensitivity to initialvalues could be resolved based on those methods.At the stage of the identification of acoustic emission source, this paper using thesample center fuzzy membership of cluster center to decide data classification. Sampledata center was optimized by differential evolution algorithm. The feasibility of thismethod was proved through experiments.The establishment of the laboratory experiment of acoustic emission source collectiondevice systems were extracted pitting, cracks and corrosion oxide film three kinds ofacoustic emission signals, this paper use clear fuzzy clustering optimization algorithm toidentify the establishment of acoustic emission source identification system model forreservoir tank bottom spot was detected acoustic emission signals, the results show that themodel can effectively identify the type of tank bottom corrosion injury exists.