ABSTRACT
The shear wave velocity (Vs) estimated using non-invasive multi-channel analysis of surface waves (MASW) tests is most useful information for site investigation. Various optimization algorithms (PSO, ABCO and TLBO) were applied to dispersion curves of MASW datasets to estimate the optimized Vs profiles. Depth and Vs prediction of each soil layer using statistical analysis based on probability density function (PDF) can increase reliability of optimized Vs profiles. A total of 100 iteration samples were generated at 30th iteration from these optimization algorithms. The PDF of converged solutions was calculated using a normal distribution with mean (μ) and standard deviation (σ) to find best-fit PDF from the histograms. The optimal thickness range for each soil layer was determined based on high-probability areas in the histograms by under- and over-estimating the mean (μ) thickness values by 2σ and mean Vs values by 5σ. The accuracy of the optimized soil layer thickness and Vs from the TLBO algorithm was successfully justified with PDF-based statistical analysis of converged iteration samples.
