| Magnetic Resonance Imaging, MRI, is one of the most valuable and promising medical dignosis technologies. As a high medical diagnostic equipment, the main magnet member of MRI can be classified as three types, i.e. permanent magnet, magnet excited by coils and superconducting magnet. In China, MRI device with permanent magnet is widely used in clinic because of its open magnet structure and low price due to the rich resource of rare-earth. However, in recent years, owing to the limited resource for permanent magnetic material, the advantage of permanent magnetic low-field MRI device is gradually disappearing. Thus superconducting MRI attracted people's attention once again. Traditional superconducting MRI device usually adopts solenoid-type structure, in which the imaging space is a cylindrical cavity with low degree of openness.To improve the performance of"openness"for the magnet structure of MRI device, a novel open-type superconducting MRI magnet structure is proposed in this thesis. Its magnet configuration is similar to one of permanent MRI device and its main magnetic field is produced by superconducting coils.In order to achieve desired design objectives including high magnetic field strength and high homogeneity in the imaging region, a global optimization algorithm is developed. Firstly, the orthogonal design (OD) was employed to define the key ones among several design variables having great effect on magnetic characteristics in the imaging region.Then, Latin hypercube sampling strategy is developed to obtain the sampling data in the design variable (DV) space. The response surface model (RSM) with multi-quadric (MQ) raial basis function is used to approximate the optimization objectives. In RSM, the selection method for the shape parameter is discussed. Finally, the evolution strategy (EA) is applied to find the global optimal solution of the RSM.In this thesis, in order to verify the validity of proposed optimization strategy, an analytic function is tested. At the same time, the TEAM standard problem 22 which is a benchmark problem to check the optimization algorithm is calculated and compared with other algorithms.Finally, the proposed algorithm is applied to optimal design of main magnet for the open-type superconducting MRI, in which the pole shape of magnet is chosen as design variables. The optimum design of magnet configuration is obtained with desired magnetic field strength and high homogeneity. |
