Fig 1.
Corrosion test set-up, connected to AE sensor and potentiostat for acquisition.
Fig 2.
Carbon-steel LPR test experimental system.
Fig 3.
Schematic flow diagram of LPR test during AE data acquisition.
Table 1.
AE parameters.
Table 2.
Specifications of R1.5I-AST sensor.
Fig 4.
A flowchart of the proposed AE features extraction and classification approach.
Fig 5.
Bias original and unbiased shifted AE signals.
Fig 6.
Different types of AE features.
Fig 7.
Five-layers wavelet packet decomposition tree.
Fig 8.
Clustering processes of Support Vector Classifier (SVC).
Fig 9.
Categorization of severity levels of uniform corrosion.
Fig 10.
Feature extraction process of AE datasets for machine learning model.
Table 3.
An example of AE features set of three domains for single region.
Table 4.
A summary of L-SVC model specifications utilized for simulation.
Table 5.
Confusion matrix utilized in our study.
Fig 11.
Specimen sample before and after LPR test.
Fig 12.
Feature heatmap of utilized AE dataset.
Fig 13.
Feature importance of utilized AE dataset.
Table 6.
ANOVA test results for the extracted features.
Fig 14.
Mean frequency power, frequency-domain features for three stages of corrosion.
Fig 15.
Peak-to-Peak (V), time-domain features for three stages of corrosion.
Fig 16.
Standard deviation, statistical features for three stages of corrosion.
Table 7.
Classification performance comparison between adopted L-SVC and benchmarked models.
Table 8.
Classification accuracy associated with error comparison between adopted L-SVC and benchmarked models.
Fig 17.
Confusion matrix of the adopted L-SVC.