Implementation of Machine Learning Algorithms in Spectral Analysis of Surface Waves (SASW) Inversion

• DOI: 10.3390/app11062557
• Publication Date: 2021-03-12T16:56:55Z
• Authors: Sadia Mannan Mitu; Norinah Abd. Rahman; Khairul Anuar Moh...; Mohd Asyraf Zulki...; Sri Atmaja P. Ros...
• Keywords: Technology; spectral analysis of surface wave; QH301-705.5; T; Physics; QC1-999; 0211 other engineering and technologies; 02 engineering and technology; Engineering (General). Civil engineering (General); inversion; Chemistry; machine learning; TA1-2040; Biology (General); QD1-999; automation

One of the complex processes in spectral analysis surface waves (SASW) data is inversion procedure. An initial soil profile needs to be assumed at beginning analysis, which involves calculating theoretical dispersion curve. If assumption starting model not reasonably close, iteration process might lead nonconvergence or take too long converged. Automating procedure will allow us evaluate stiffness properties conveniently and rapidly by means SASW method. Multilayer perceptron (MLP), random forest (RF), support vector regression (SVR), linear (LR) algorithms were implemented order automate inversion. For this purpose, curves obtained from 50 field tests used as input for all algorithms. The results illustrated that SVR could potentially estimate shear wave velocity soil.