Research On Atmospheric Storage Tank Bottom Plate Corrosion Characteristic Identification Based On Acoustic Emission

Atmospheric storage tank bottom corrosion process is a very complex problem.It involves large area of uniform corrosion,localized pitiing corrosion and special position stress corrosion which appears on the tank floors,even coupling with several types of corrosion when it becomes more serious.Tank bottom corrosion and perforation leakage are a serious threat to the operational safety of the storage tank.Although on-line acoustic emission testing(AET)of tank bottom corrosion can grade the tank floor by “good tank” and “bad tank” according to testing results.The qualitative identification of corrosion AE source is the bottleneck of AE inspection technology.For a long time,there have been many difficulties in improving the accuracy of corrosion identification and determination of the best interval of internal inspection.Therefore,it is very important to do some research on tank bottom corrosion characteristics identification using AE technology.In this dissertation,the AE source mechanisms of uniform corrosion,pitting corrosion and stress corrosion of tank bottom steel were studied using electrochemical polarization and AE method.The AE characteristic parameters analysis,frequency spectrum analysis and wavelet transform(WT)were used to process the AE signal and extract the features of corrosion AE source,further using the corrosion type identification diagrams and the improved BP neutral network to do pattern recognition of the AE source.The main research work and the results are as follows:1.The AE signal features of uniform corrosion of tank bottom steel were studied by AE method and electrochemical method.The results showed that the main AE sources in uniform corrosion were corrosion itself and corrosion products activity.They both generated low-frequency AE signal,the frequency was mainly concentrated in the region of 20~70 kHz.Uniform corrosion itself mainly generated low-amplitude and low-duration signals,while corrosion products activity generated relatively higher amplitude and duration signals.2.In the pitting corrosion AE monitoring experiments,it found that hydrogen bubble oscillation and breakage,oxided film rupture,corrosion products layer friction,pits growth and propagation released AE signal.The low cut-off frequency of hydrogen bubble breakage signal was about 130 kHz validated by Minnaert bubbling frequency equation,while both the pits growth signal and oxide film rupture distribution range was main between 80 to 130 kHz.The overwhelmingly dominant pits growth signals had a low duration less than 80?s,while the oxide film rupture signal duration was generally greater than 100?s.3.In the stress corrosion AE monitoring experiments,it was confirmed that the micro-crack initiation,macro-crack propagation and subsidiary pitting were the main corrosion AE source,which produce high frequency AE signals.The micro-crack signal took obvious advantage and generally throughout the whole acquisition time,its duration was relative divergence,and its frequency was concentrated in the region of 110~190 kHz,while macro-crack propagation signal appeared in the last stage of the experiment with high counts number,great energy and relatively concentrated duration,with frequency distribution range mainly between 260~400 kHz of high frequency band.4.A discrimination diagram of corrosion types was proposed based on AE signal frequency-duration identification and frequency-counts identification by extracting the feature of corrosion AE source,and it can be used to identify the tank bottom corrosion primarily and intuitively.A BP neural network was also built based on the 24-dimension feature vectors formed by the AE characteristic parameters and the wavelet package characteristic energy spectrum coefficients,extracted from time domain,frequency domain and energy domain of the signal.The identification results were analyzed by the fuzzy matrix.It was confirmed that the improved BP network can realize the pattern recognition of uniform corrosion,pitting corrosion and stress corrosion with high accuracy and good stability.