中文摘要：

针对永磁驱动器（PMD）的结构设计问题,提出一种基于改进熵权法结合混合代理模型的优化设计方法。首先利用基于交叉验证误差的最优加权法,将响应曲面法、克里金法以及支持向量机回归结合起来,构建PMD的参数变量与响应变量之间的混合代理模型;然后引入改进的熵权法,将PMD的多指标转化为单一综合指标,并建立其优化的数学模型,通过自适应权重粒子群优化算法求解;最后对结果进行有限元仿真分析和实验室仿真平台验证。研究结果表明,所提出的优化设计方法优于其它方法,得到的PMD结构参数合理有效,较好的实现了PMD的多目标优化设计。

英文摘要：

Permanent magnet drive （PMD） design problems were studied and an optimization design approach based on hybrid surrogate model and improved entropy-weight was put forward. Firstly by combining the response surface method （RSM）, Kriging method and support vector machine regression （SVR） , a hybrid surrogate model of optimizing parameters and response variables was built, in which the optimal weighting method based on cross validation error is used; Secondly the improved entropy-weight was introduced to transform the multi-indexes of PMD into a single composite one, and the mathematical model was built, which was solved by adaptive weight particle swarm optimization particle swarm optimization （AWPSO） ; Finally the parameters results were analyzed with finite element method and experimental validation. The results show that the proposed method is superior to other methods, and the structural parameters are reasonable and effective. And the expected results are derived.