在進(jìn)行MRI （magnetic resonance imaging）超導(dǎo)主磁體的設(shè)計(jì)時(shí)常采用優(yōu)化設(shè)計(jì)的方法，將各設(shè)計(jì)參數(shù)看作連續(xù)變量處理，但實(shí)際上很多參數(shù)是離散變量，為了更符合工程實(shí)際，將超導(dǎo)MRI 主磁體的設(shè)計(jì)作為一個(gè)含有離散變量的全局優(yōu)化問題。建立了適用于多種超導(dǎo)MRI 主磁體結(jié)構(gòu)的數(shù)學(xué)模型，包括設(shè)計(jì)變量、目標(biāo)函數(shù)、約束條件等，選用了適用于MRI 超導(dǎo)主磁體優(yōu)化設(shè)計(jì)的含有離散變量的模擬退火算法進(jìn)行設(shè)計(jì)。算例結(jié)果表明，本文選取的數(shù)學(xué)模型和優(yōu)化算法是有效的，能夠達(dá)到超導(dǎo)MRI 主磁體設(shè)計(jì)的要求。
關(guān)鍵字：磁共振成像；超導(dǎo)主磁體；離散優(yōu)化；模擬退火
Abstract: Optimization design is often used in the design of MRI superconducting main magnet, where all variables are treated as continuous ones. According to the fact that not all variables are continuous, here the design problem is seen as a global optimization problem with discrete variables. A mathematical model is built including the definition of variables, objective function and constraints. A simulated annealing algorithm suitable for the
optimization of mixed integer non-linear programming problems is used to design MRI SC main magnet. An example is given to illustrate the effectiveness of the mathematical model and the optimization algorithm.
Key words: MRI; superconducting main magnet; discrete optimization; simulated annealing