Generic probabilistic inversion technique for geotechnical and transportation engineering applications

A wide range of important problems in civil engineering can be classified as inverse problems. In such problems, the observational data related to the performance of a system is known, and the characteristics of the system or the input are sought. There are two general approaches to the solution of inverse problems: deterministic and probabilistic. Traditionally, inverse problems in civil engineering have been solved using a deterministic approach. In this approach, the objective is to find a model of the system that its theoretical response best fits the observed data. In deterministic approach to the solution of inverse problems, it is implicitly assumed that the uncertainties in data and theoretical models are negligible. However, this assumption is not valid in many applications, and therefore, effects of data and modeling uncertainties on the obtained solution should be evaluated. In this dissertation, a general probabilistic approach to the solution of the inverse problems is introduced, which offers the framework required to obtain uncertainty measures for the solution. Techniques for direct analytical evaluation and numerical approximation of the probabilistic solution using Monte Carlo Markov Chains (MCMC), with and without Neighborhood Algorithm (NA) approximation, are introduced and explained. The application of the presented concepts and techniques are then illustrated for three important classes of inverse problems in geotechnical and transportation engineering as application examples. These applications are: Falling Weight Deflectometer (FWD) backcalculation, model calibration based on geotechnical instrument measurements, and seismic waveform inversion for shallow subsurface characterization. For each application, the probabilistic formulation is presented; the solution is obtained; and the advantages of the probabilistic approach are illustrated and discussed.Ph. D.Includes bibliographical references