Cluster analysis of acoustic emission signals in pitting corrosion of low carbon steel

Abstract

The pitting corrosion characteristics of low carbon steel specimens are studied by acoustic emission (AE) and electrochemical techniques, in a 3.0 wt.% NaCl solution acidified to pH 2.0. The acoustic emission signals generated by pitting corrosion are classified based on multiple acoustic emission parameters using K-means clustering algorithm, then each classified signals are analyzed by acoustic emission parameters correlation plot and distribution with time. Furthermore, each acoustic source characteristics is extracted using Gabor wavelet transform (WT) in the time and frequency domain. An error back propagation (BP) artificial neural network (ANN) is trained according to the classified signals, so as to successfully identify the acoustic emission signals from parallel experiments. Experimental results show that the hydrogen bubble activation, oxidized film rupture and pit growth are typical acoustic emission sources in pitting corrosion process, which can be effectively classified by cluster analysis and recognized by back propagation neural network. The data gathered from laboratory tests combined with the real data from acoustic emission online storage tank floor inspection can help to evaluate the bottom corrosion severity and interpreter the corrosion source, further to make the on-site testing more reliable and reduce the risk.

Pitting Judgment Model Based on Machine Learning and Feature Optimization Methods

“…Due to its advantages in dealing with multi-dimensional, nonlinear and uncertain characteristics, machine learning (ML) methods have been gradually applied in the field of corrosion science in recent years (Hu et al, 2014;Bi et al, 2015), and have been successfully applied in some pitting corrosion related simulations. The pitting corrosion prediction model based on ML can not only describe the nonlinear relationship between the influencing factors and the target parameters, so as to realize the accurate prediction of the pitting information, but also can effectively extract the important feature information that reflects the health state of steel in the corrosion data (Diao et al, 2021).…”

Acoustic Emission of Metallic Specimen with Surface Defect During Fatigue Crack Growth

“…Previous studies on AE were focusing on hit-related information during fatigue tests, including the relationship between count rate and material rolling direction [24], and the relationship between count number on fatigue cycle and crack propagation rate [5]. Nevertheless, recent works on AE have focused on waveform pattern analysis and highlighted the importance of studying AE waveform [25,26], because it could provide more information than only analysing hit-related characteristics. It is found that AE signals of aluminium alloys 7075-T6 have peak frequencies around 100 kHz, 260 kHz and 600 kHz [20].…”

“…It is found that AE signals of aluminium alloys 7075-T6 have peak frequencies around 100 kHz, 260 kHz and 600 kHz [20]. Other research also found different frequency peaks of low carbon steel [26] or aluminium alloy 2024 T3 [27]. Studies on AE waveforms are often challenging because the amplitude and frequency of AE waveforms can vary significantly.…”

“…The group velocities for distinct wave modes vary with frequency due to its dispersive nature and can be used in identifying wave modes. The most frequently used AE sensors are piezoelectric wafer active sensors and typically are threshold-based sensors that only record AE signals which exceed the set threshold level to discard possible background noise [26]. Unlike the active wave sensing approach, where guided waves are excited by a transmitter sensor [29], AE is a passive monitoring approach that listens for waves generated by rapid energy released from the structures themselves.…”

Acoustic emission signal processing framework to identify fracture in aluminum alloys

“…Furthermore, machine learning tools have been also used to correlate AE content with fracture and damage in general. Specifically, both unsupervised [39,[56][57][58] and supervised [30] clustering algorithms are commonly used to separate noise from damage sources using features that are selected to achieve the desired clustering. Among these methods, k-means clustering is the most commonly used to separate AE signals based on feature characteristics.…”