IAES International Journal of Artificial Intelligence (IJ-AI) Vol. 13, No. 2, June 2024, pp. 1459~1468 ISSN: 2252-8938, DOI: 10.11591/ijai.v13.i2.pp1459-1468  1459 Journal homepage: http://ijai.iaescore.com Towards an optimization of automatic defect detection by artificial neural network using Lamb waves Nissabouri Salah 1 , Elhadji Barra Ndiaye 2 1 M2S2I Laboratory, Department of Mechanical engineering, ENSET-Mohammedia, Hassan II University, Mohammedia, Morocco 2 Aerospace Technology Center, Montreal, Canada Article Info ABSTRACT Article history: Received May 9, 2023 Revised Oct 6, 2023 Accepted Oct 21, 2023 This paper presents a damage detection method based on the inverse pattern recognition technique by artificial neural network (ANN) using ultrasonic waves. Lamb waves are guided elastic waves, are widely employed in nondestructive testing thanks to their attractive properties such as their sensitivity to the small defects. In this work, finite element method was conducted by Abaqus to study Lamb modes propagation. A data collection is performed by the signals recorded from the sensor of 300 models: healthy and damaged plates excited by a tone burst signal with the frequencies: 100 kHz, 125 kHz, 150 kHz, 175 kHz, 200 kHz, and 225 kHz. The captured signals in undamaged plat are the baseline, whereas the signals measured in damaged plates are recorded for various positions of external rectangular defects. To reduce the amount of training data, only two peaks of measured signals are required to be the input of the model. Continuous wavelet transform (CWT) was adopted to calculate the key features of the signal in the time domain. The feed forward neural network is implemented using MATLAB program. The data are divided as follows: 70% for training the model, 25% for the validation, and 5% for the test. The proposed model is accurate estimating the position of the defect with an accuracy of 99.98%. Keywords: Artificial neural network Continuous wavelet transform Damage detectio Lamb waves Ultrasonic waves This is an open access article under the CC BY-SA license. Corresponding Author: Nissabouri Salah M2S2I Laboratory, Department of Mechanical engineering, ENSET-Mohammedia, Hassan II University Mohammedia, Morocco Email: salah.nissabouri@univh2c.ma 1. INTRODUCTION The mechanical parts are often subjected to many constraints and deformation leading to the appearance of external defect. This causes the weaknesses or their fracture. Therefore, many nondestructive testing methods are developed to predict the defect even before the first stage. One of the most used methods is Lamb waves. These waves are sensitive to small defect. They are used to control the corrosion of pipe [1] welding defect [2] in isotropic plate [3] or in composite cylinders [4]. One of the challenging tasks is to choose the adequate mode and the corresponding signals. Lamb waves consisted of propagating of fundamentals modes: antisymmetric noted A0 and S0. When the signal interacts with the defect, the propagating mode may convert to the other mode. This variation of the nature can noise the post processing and induce error detection. So many authors have studied the conversion phenomenon like [5], [6]. Post processing is a crucial step where the characteristics of captured signals are calculated. Among many, continuous wavelet transform (CWT) is most used in the field of nondestructive testing. It allows the analysis of time-frequency information, extracts and separates out frequency information from a timeseries.