一种基于差分进化理论优化的隐式空间映射算法及其应用

摘要

False convergence may exists in filter optimisation process of standard implicit space mapping algorithm,it leads to the mapping relationship between coarse-model and fine-model becoming weak and makes the entire optimisation efficiency be low.We propose an improved algorithm,by introducing differential evolution theory it searches the optimal global solution of objective function,and prevents false conver-gence in optimisation process.At last,we apply this algorithm to optimise an LTCC filter with multi-level structure.Experimental results dem-onstrate that the new method aimed at achieving better mapping relationship has less iteration times and costs shorter time under the condition of satisfying the design specifications similarly.It has good feasibility and efficiency.%标准隐式空间映射算法在优化滤波器的过程中可能会出现假收敛现象，从而导致模型之间建立的映射关系偏弱，使得整体的优化效率降低。介绍一种改进算法，通过引入差分进化算法（DE）搜索目标函数的全局最优解，避免优化过程出现假收敛。最后应用该算法优化一种多层结构的 LTCC 微带滤波器。实验结果表明，旨在获取更好映射关系的新算法在同样满足设计指标的条件下，所用迭代次数少，用时也更短，具备良好的可行性和高效性。