Machine Learning-Based Corrosion-Like Defect Estimation With Shear-Horizontal Guided Waves Improved by Mode Separation

Abstract

Shear Horizontal (SH) guided waves have been extensively used to estimate and detect defects in structures like plates and pipes. Depending on the frequency and plate thickness, more than one guided-wave mode propagates, which renders signal interpretation complicated due to mode mixing and complex behavior of each individual mode interacting with defects. This paper investigates the use of machine learning models to analyse the two lowest order SH guided modes, for quantitative size estimation and detection of corrosion-like defects in aluminium plates. The main contribution of the present work is to show that mode separation through machine learning improves the effectiveness of predictive models. Numerical simulations have been performed to generate time series for creating the estimators, while experimental data have been used to validate them. We show that a full mode separation scheme decreased the error rate of the final model by 30% and 67% in defect size estimation and detection respectively.

A pipeline defect estimation method based on supervised learning fusion model

“…(1) Qualitative model selection and optimization validation A selection of various hyper-parameter values will impact the model prediction performance. After testing, there are two crucial hyper-parameters in XGBoost to be considered: max_depth and n_estimators, where max_depth denotes the splitting depth, set its initial value range from 5 to 20 in steps of 2; n_estimators denotes the number of trees, set its initial value to the interval of (300, 450), and finally, through the grid The optimal parameter search interval is determined as (18,20)(400, 410) respectively by the grid search method. After dividing the data set according to step 1, the model was tuned to the best state according to the tuning process given in step 2, and Table 4 shows the comparison of each model parameter before and after optimization.…”

“…Chen, Herzer et al [17] investigated parameter-optimized machine learning algorithms for defect classification and demonstrated that hyperparametric search-based SVM and RF algorithms can obtain the highest classification accuracy. STEVE DIXON [18] proposed a machine learning-based shear level guided wave corrosion class defect estimation, which effectively applied the supervised learning method to the defect estimation of SH-guided wave. Although the computability can be improved by the prediction model, it only used the statistical characteristics from other health monitoring fields for the extraction of statistical features, and cannot be used as a typical feature representation of ultrasonic defect signals, while the estimation rate of the crack, hole, and broken slope are 0.63, 0.88 and 0.90.…”

Detection of weld defects using ultrasonic-guided waves based on matching pursuit and density peak clustering

“…Amer and Kopsaftopoulos (2022) used the statistical mathematical model to detect the defects of carbon fiber reinforced plastic (CFRP) plate by guided wave, but they did not identify the mode of guided wave signal. The machine learning model is also used to separate shear horizontal (SH) wave modes (De Castro Ribeiro et al, 2021), but only the two lowest SH order-guided modes of SH0 and SH1 are analyzed.…”

Hybrid Entropy in the Time-Frequency Domain for Grading Electrode Sediment Identification

“…To achieve nondestructive detection of pipe defects, numerous studies have focused on the use of electromagnetic acoustic transducers (EMATs) [8], circumferential Lamb waves [9], and ultrasonic guided waves [10]. However, the above methods can only detect defects on or near the surface of the workpiece, and are no longer applicable to the inspection of grading electrodes located inside composite pipes.…”