The Design And Optimization Of Permanent Magnet Structure Producing A Slice Of Homogeneous Magnetic Field

The permanent magnet Magnetic Resonance Imaging (MRI) device is a kind of advanced medical diagnostic device, which can obtain the images of interior tissue of human body. Due to its good openness and lower maintenance fee, the completely open permanent magnet MRI is attracting more and more people's attention. The main magnet is one of the main parts of MRI. The permanent magnet structure involved in this paper, which can produce a slice of homogeneous magnetic field, is just about to use as the main magnet of MRI. Aimed at how to obtain a better main magnet structure, this paper has done the following research work:Firstly, on the basis of introducing the theory of Response Surface Methodology (RSM), the advantages and disadvantages of several kinds of interpolation basis functions used to construct response surface model in present has been summarized, and how to assume the shape parameter a value when using radial basis function as the interpolation function has been discussed too. Moreover RSM and the improved Genetic Algorithm (GA) have been combined together, and a new kind of optimization method which can solve large magnetic field inverse problems has been proposed, and was fulfilled by programming.Secondly, the advantages and disadvantages of classical multi-objective optimization method has been analyzed, the multi-objective optimization method based on GA has been detailedly introduced. Furthermore the typical non-Pareto optimization method--vector evaluated GA, the optimization method based on Pareto—Non-dominated Sorting GA (NSGA) and the fast elitist Non-dominated Sorting GA (NSGA-Ⅱ) have been mainly discussed. Moreover, the RSM was introduced into the NSGA-Ⅱ, which makes the multi-objective optimization of large magnetic field inverse problem more efficient and timesaving. At the same time aimed at the characteristic of the populations of the NSGA-Ⅱnot disperse enough, the original crossover operator of the NSGA-Ⅱhas been improved. Numerical experiment results show the improved method is very efficient. At last, a kind of permanent magnet structure which can produce a slice of homogeneous magnetic field has been presented, and a process of single-objective optimization has been fulfilled. After optimization the results have been analyzed by electromagnetic field analytical tools, and it has been proved that the magnetic field flux density, the homogeneity of each layer and the gradient between each layer could achieve the design requirements. This makes the model can use as a main magnet of MRI. Furthermore, a multi-objective optimization process of the permanent magnet structure using the proposed method has been completed, and the results can satisfy different design requirements. And a 3D model based on real model has been created by ANSOFT, comparison between f'mite element analytical results and accurate experimental results show the analytical tool is valid.