TITLE:
Cone Bearing Estimation Utilizing a Hybrid HMM and IFM Smoother Filter Formulation
AUTHORS:
Erick Baziw, Gerald Verbeek
KEYWORDS:
Bayesian Recursive Estimation (BRE), Cone Penetration Testing (CPT), Geotechnical Site Characterization, Hidden Markov Model (HMM), Iterative Forward Modelling (IFM), Smoothing
JOURNAL NAME:
International Journal of Geosciences,
Vol.12 No.11,
November
18,
2021
ABSTRACT: Cone penetration testing (CPT) is a widely used
geotechnical engineering in-situ test
for mapping soil profiles and assessing soil properties. In CPT, a cone on the
end of a series of rods is pushed into the ground at a constant rate and
resistance to the cone tip is measured (qm). The qm values are utilized to characterize the soil profile. Unfortunately, the
measured cone tip resistance is blurred and/or averaged which can result in the distortion of the soil
profile characterization and the inability to identify thin layers. This paper
outlines a novel and highly effective algorithm for obtaining cone bearing
estimates qt from averaged or smoothed qm measurements. This qt optimal filter estimation technique is
referred to as the qtHMM-IFM algorithm and it implements a
hybrid hidden Markov model and iterative forward modelling technique. The
mathematical details of the qtHMM-IFM algorithm are outlined in this paper along with the results from
challenging test bed. The test bed simulations have demonstrated that the qtHMM-IFM algorithm can derive accurate qt values from challenging
averaged qm profiles. This allows for greater soil resolution
and the identification and quantification of thin layers in a soil profile.