Search

This paper presents a nature-based algorithm, titled multivariable teaching-learning-based optimization (MTLBO) algorithm. MTLBO algorithm during an iterative process can estimates the best values of the buried structure (model) parameters in a multi-objective problem. The algorithm works in two computa-tional phases: the teacher phase and the learner phase. The major purpose of the MTLBO algorithm is to modify the value of the learners and thus, improving the value of the model parameters which leads to the optimal solution. The variables of each learner (model) are the depth (z), amplitude coefficient (k), shape factor (q), angle of effective magnetization (Î,) and axis location (x0) parameters. We employ MTLBO method for the magnetic anomalies caused by the buried structures with a simple geometric shape such as sphere and horizontal cylinder. The efficiency of the MTLBO is also studied by noise corruption synthetic data, as the acceptable results were obtained. We have applied the MTLBO for the interpretation of the four magnetic anomaly profiles from Iran, Brazil and India. [ABSTRACT FROM AUTHOR]

The paper describes some recent applications of the measurements of ambient vibrations (micro seismic noise) in Croatia. They include free-field measurements, as well as those done within buildings. Data obtained in the field at the studied localities are consistent with the properties of shallow geological structures known to exist there. In Zagreb, Horizontal-to-Vertical Spectral Ratios (HVSR) indicate thick alluvial cover (over 100 m) that gradually gets thinner towards the slopes of the Mt. Medvednica. A similar situation-on a smaller scale-is also encountered in Ston, where HVSR profiles reveal several tens of meters thick sedimentary cover over the bedrock which gets exposed at the Bartolomija hill. Analyses of records from Ston and Dubrovnik suggest that soil-building resonance must be seriously considered. Measurements in the buildings were analyzed by a newly developed software package FREDA-a suite of Matlab routines. Based on tests on synthetic and real data it is concluded that spectral methods are in general more robust and less dependent on parameters of the algorithm employed, than time-domain analyses. Spectral algorithms are also better in resolving higher modes. Applications to real buildings proved that analyses of shaking induced by ambient vibrations in most cases lead to well constrained, reliable, and time independent estimates of frequencies and damping of the buildings' vibrational modes for small excitation levels. [ABSTRACT FROM AUTHOR]

The implementation of VGR-stress imaging technique, as a free surface boundary condition, in a (2, 4) staggered grid P-SV wave finite difference (FD) algorithm with a variable size of a grid cell is presented in this paper. VGR is acronym for 'vertical grid-size reduction'. The qualitative and quantitative results confirmed that the effective thickness (ETH) of the first soil layer become less by one-half of the vertical size of a grid cell than the assigned thickness (ATH), if stress imaging technique is used as a free surface boundary condition. The results of various numerical experiments revealed that the stress imaging technique causes significant numerical dispersion of Rayleigh waves and the VGR-stress imaging technique is efficient enough to avoid the same. So, superiority associated with the VGR-stress imaging technique over the well stress imaging technique is that it avoids both the significant numerical dispersion of Rayleigh waves in homogeneous medium and the soil thickness discrepancy. The maximum grid spacing ratio (ratio of largest to smallest size of a grid cell) up to 6.0 did not affect the accuracy of FD algorithm with a variable size of a grid cell. In case of a variable size of a grid cell, the required computational memory and time for a particular basin-edge model was 6.43 and 16.62 times lesser than that required in case of uniform grid. [ABSTRACT FROM AUTHOR]