Parallel Genetic Algorithms And Its Application To The Optimization Design Of Main Magnet Assembly Of MRI

Nuclear magnetic resonance imaging or magnetic resonance imaging (MRI) is a kind of imaging technology of biological magnetic nuclear spin and the most developing imaging technology, which depended on the development of computer, electro and superconductivity. Main magnet body is one of the main parts of MRI. It can product magnetic field needed by MRI. The device of MRI adopts permanent magnet main magnet in this paper.Optimization design of main magnet belongs to inversion problem of electromagnetism field. The calculating time of inversion problem of electromagnetism field is very long. Using parallel disposal technology can reduce computing time effectively. With the development of parallel disposal technology, parallel workstation had become mainstream of parallel calculating. Inherent parallel nature of Genetic Algorithm and fast development of parallel computers, so many scientists begin studying parallel problem of Genetic Algorithm.This paper introduces Parallel Genetic Algorithm to optimize the main magnet of MRI. First, simply analyze Genetic Algorithm and its running mechanism. Changing Genetic Algorithm to parallel mode and forming Parallel Genetic Algorithm. Introduce modes of Parallel Genetic Algorithm. Because this paper adopts thick granularity mode, introduce thick granularity mode mostly.Second, analyzing basic theory of MRI and analyzing the influence of pole shape of MRI in imaging area. This paper optimizes the size of pole shape. Using Parallel Genetic Algorithm and finite element analysis, make optimization realize by one computer.Third, introducing parallel software called PVM, which can help actualizing Parallel Genetic Algorithm. Parallel Virtual Machine also called PVM can distribute computing. It is a software package. Using it can change computers to a big parallel computer by network. If doing this, we can spend little time completing task of parallel optimization. This paper adopts PVM to realize Parallel Genetic Algorithms. Parallel flat roof is made up of Parallel Virtual Machine (PVM) and Remote Shell Daemon (RSHD/NT). Basing this parallel flat roof, we can accomplish the task of parallel optimization by using program of Genetic Algorithms. Results indicate that using Parallel Genetic Algorithm can enhance calculating speed and reduce calculating time.At last, deficiency is given about the whole paper and the further researches are proposed.