基于代理模型的直线感应电机多目标优化设计

项目名称： 基于代理模型的直线感应电机多目标优化设计

项目编号： No.51307099

项目类型： 青年科学基金项目

立项/批准年度： 2014

项目学科： 电工技术

项目作者： 宫金林

作者单位： 山东大学

项目金额： 25万元

中文摘要： 本项目旨在综合运用有限元分析方法、代理模型法、多目标优化理论等知识设计直线感应电机。具体内容包括：（1）建立以电磁学模型和热力学模型联合仿真的多物理有限元参数化模型，提高对直线电机的分析能力并对其进行基本的理论研究，为进一步的优化设计提供精确模型；（2）在有限元模型的基础上，建立代理模型并对优化问题进行分析，筛选优化参数，确定优化问题；（3）基于代理模型的多目标优化策略的研究，其中包括代理模型引导采样的多目标优化策略和多级多目标空间映射的优化策略。充分利用代理模型的分析速度和有限元模型的计算精度，为电机的设计提供经过有限元模型评估的Pareto前沿解集。本项目的研究为直线感应电机的多目标优化设计奠定坚实的理论基础和技术基础。

中文关键词： 代理模型；有限元模型；有效全局优化；空间映射优化；

英文摘要： The purpose of this project is to integrate the finite element analysis method, surrogate-assisted multi-objective optimization theory to conceive a linear induction motor. Concrete contents include: (1) a multi-physical model which consists of a 3D finite element electromagnetic model and a thermal one are built and parametered. The analysis capacity of the linear motor is improved and the basic theory is studied using the multi-physique model, which can supply an accurate model for the optimization;(2)Using the finite element model, the surrogate model is built and then used to analyse the optimization problem, the optimization parameters are screened and the final optimization problem is formulted;(3)The surrogate-assisted multi-objective optimization strategies are studied, including the multi-objective optimization method based on the surrogate model guiding samples and the multi-leve and multi-objective space mapping optimization method. The proposed strategies benefit both the rapidity of the surrogate model and the precison of the finite element model. A Pareto front solution composed of the finite element model evaluation results is proposed, which allows taking the decision for the motor design. In this projet, we provide the solid basic theory and technique for the multi-objective optimization of a li

英文关键词： surrogate model；finite element model；efficient global optimization；space mapping technique；