This dissertation explores different aspects of the detection of hidden objects, including its theoretical modelling and numerical investigation. First, the magnetic properties of layered soil are analyzed using a simulation package called HysterSoft. It is a program for simulating magnetic material hysteresis phenomena. To combine phenomenological models of hysteresis, it offers a user-friendly interface. The Energetic model, the Hodgdon model, the Jiles-Atherton model, the Langevin model, the Preisach model, and the Bouc-Wen model are some of the hysteresis models that have been implemented thus far. The nature of soil accounts primarily for the level of coercive field it may have. Low coercive field of hysterons in soils leads to high magnetization, permittivity, and permeability. High coercive field of hysterons in soils leads to low magnetization, permittivity, and permeability. The properties of soils are considered in the modeling of hidden objects detection by electromagnetics method.This dissertation uses soil model with specific permittivity and permeability to model the environment where a hidden object to be detected is determined. The feature of the hidden object is extracted based on its Radar Cross Section (RCS) feature. Furthermore, using the full wave electromagnetic modeling program FEKO, we used a layered soil model to describe subsurface objects, particularly concealed/hidden objects. The name FEKO is from the German acronym "Feldberechnung für K?rper mit beliebiger Oberfl?che," which means "field calculations with bodies of arbitrary shape." It is used as 3D electromagnetic simulator. By adjusting the material characteristics of soil based on permittivity and conductivity of the surrounding soils, depth, and incoming incident wave direction for detection of the buried item, we undertake a parametric investigation of the radar cross section (RCS) characteristics. Several simulations results shed light on how the physical characteristics of the examined hidden object and its radar signatures are related. The technique used has the benefit of allowing for the efficient collection of hidden object's RCS data with consideration to surrounding soils.Furthermore, we examined the Gauss-Newton based algorithm in this dissertation by using the algorithm uniquely for electromagnetic nonlinear hysteresis curve. Afterward, we modified the Gauss-Newton based algorithm to achieve a more efficient and robust performance by factoring noise into the algorithm. We apply this robust algorithm to the Generalized Prandtl Ishilinskii. Thereafter, a modified GPI (mGPI) is developed for a better Gauss-Newton iterative convergence. Nonlinear transformations are used on this Gauss-Newton based algorithm.The unknown model parameters for GPI must be re-defined so that they fall inside the permitted physical range i.e., symmetrical about the origin. Using the least square technique and linear search methods, we modified the iterative stage to improve the algorithm's error-reducing characteristics. The development of our modified Gauss-Newton method demonstrates that by characterizing our cost function as a multiplicative function, we determined that the unknown parameters may be identified with confidence limits specified by an established error limit.
展开▼
