Characteristics Study Of The Tank Bottom Corrosion Process Acoudtic Emission Signals Based On Generic Algorithm

Acoustic emission (AE) detection technology, is a kind of continuous detection, thereare not obvious requirements of testing environment and the member shape. It can achievethe large scale detection on line. In recent years, much more attention has been obtained.Acoustic emission technology began in the1950s, in the early sixties was introduced intothe field of non-destructive testing, and then, AE gradually came to the study of pressurevessels. In recent years, AE as a new method for the detection of tank bottom, has gottenmore and more attention by people. Acoustic emission monitoring of tank floor corrosionconditions are very effective online monitoring methods, it is easy to detect early corrosionof the tank bottom.One of the main tasks of feature extraction, is to extract some useful features from thewaveform which is mixed with noise. In this paper, the Matching Pursuit algorithm is usedto select the best atom which can match the AE waveform from the over-completed atomdictionary, and then, the atom is used to reconfigure theAE signal, so the features can beextracted. However, this method will increase the decomposition difficulty of calculatingand the amount of signal, in actual operation is difficult to overcome this difficulty.For the characteristics of random and non-smooth of the tank bottomAE signals, inthis paper, the method based on genetic-matching pursuit algorithm has been used tosolve this problem. The main characteristics of the signals are given by the acousticemission detector. The Matching Pursuit (MP) arithmetic is used to extract the waveformcharacteristic. The AE signals can be well reconfigured. The genetic algorithm (GA) wasused to optimize MP algorithmic. The projection on the atom of the signal or its residue inMP arithmetic was served as the GAfitness function, and the best matching atomic parameter was confirmed. The experimental results show that the best pursuit atomicparameters are extracted and the amount of calculation is reduced substantially by thismethod. In the reconstructed signal process, signal noise can be effectively removed by thismethod, so as to achieve a certain de-noising effect. It has much practical value and a lot oftheoretical application value. At the subsequent extraction stage, the energy of corrosionsignals is also calculated, and the concept of energy distribution percentages is proposedwhich can be more intuitive to depict the active scope and intensity of corrosion AEsignals.