The Application Of Genetic Algorithm In The Dielectric Imaging

Dielectric imaging is one kind of electromagnetic imaging technologies. It reconstructs dielectric property distributions of the object exposed to electromagnetic fields by measuring fields around it. So it is an inverse problem. This thesis presents the application of genetic algorithm (GA) to reconstruct the images of object exposed to static electric field from the voltage measurements around it. The thesis includes some sections as following.Firstly, as a forward problem is solved for each iteration of GA reconstruction, calculating the electric field for a given distribution permitivity and driving source-this aspect will be treated with more detail. The thesis uses finite element method (FEM) to solve it, because FEM can conveniently handle inhomogeneous dielectric, converge well, and spare computing time by compacting the matrix.Secondly, the thesis uses the filtering method based on wavelet in signal processing. It raises a new technique to restrain drifting of modulus maxima based on ad hoc algorithm. The numeric result shows it gains good effect in searching modulus maxima lines.Thirdly, In the searching process of dielectric imaging, the thesis chooses LGA and makes improvements in the selection and crossover operators to control the course of convergence well. The selection remains the best individual if its fitness is larger than given threshold. The crossover uses a larger mutation probability if the difference between individuals is small. The numeric computation shows that those improvements can increase GA's efficiency and dielectric imaging precision.Lastly, the Windows interface of the simulation software is developed for user. It utilizes Mixed-Language Programming with FORTRAN and Visual Basic because FORTRAN is powerful for numeric computing and Visual Basic is convenient for developing user interface.