Research On Corrosion Acoustic Emission Signal Evaluation Of Tank Bottom

The safe operation of the pressure vessel problems relate to the entire petrochemicalindustry. So the pressure vessels regularly destructive testing is particularly important,through the regular nondestructive testing can timely find out defects in the storage tank. Inthe tank bottom corrosion testing, acoustic emission method compare to other traditionaldetection method has many advantages, such as real-time acquisition, continuousmonitoring and lowly testing cost. Acoustic emission method can earlier find hiddendefects existing in the pressure vessels, so as to ensure the safe operation of the pressurevessels. It is a very promising detection method.At present commonly corrosion evaluation method has weightlessness method andmechanical method, but the weightlessness method is mainly used to detect small metalspecimens in the laboratory, and machinery can cause some damage to metal. According tothe defects of the two methods, corrosion acoustic emission signal evaluation methodbased on support vector machine (SVM) is proposed. SVM for multi-dimensional dataclassification has a very good effect.This paper introduce the tank bottom corrosion detection of significance, acousticemission and acoustic emission technology development and its application of technicalprinciples, and analyze the corrosion mechanism of the acoustic emission signal, due tomulti-dimensional acoustic emission signals is difficult to deal with. SVM classificationresults only depend on the support vector, and have nothing with dimension, it is a goodway to solve this problem. So the application of support vector machine (SVM) researchedacoustic emission signal recognition effect in the lab. Radial basis function (RBF) is usedto be the kernel function of support vector machine, three kinds of parameter optimizationmethod is used to select the key parameter of RBF kernel function: The grid search method,genetic algorithm and particle swarm optimization (PSO) parameter optimization. RBFkernel function key parameters are analyzed by three kinds of parameter optimizationmethod. PC-2acoustic emission system is used to design corrosion acoustic emission signalexperimental platform. Tank commonly used metal material Q235steel make long localcorrosion experiment. Through experimental devices observe and collect signals ofdifferent corrosion stages. In the experiment the collected data such as amplitude, theimpact number and rise time establish the support vector machine (SVM) classificationmodel, and it use experimental data to test the support vector machine classification effect.Results of test show that the method has certain reference significance for tank corrosionevaluation.