基于微多相粒子群算法的介质粗糙面反演研究

摘要

A novel approach known as micro multi-phase particle swarm optimization (uMPPSO) is applied to the inverse problem associated with dielectric rough surface. A scattering model based on the method of moment (MOM) and forward and back method ( FBM) is used to solve the scattering problem. The generation of initial swarm is incorporated with the information obtained from simplex method. The error between the measured back scattering coefficient and the computed back scattering coefficient are considered as the object functions. The inverse scattering problem is transferred into an optimization problem by minimizing the object function with optimization parameters, which is solved by uMPPSO. The operation procedure is shown clearly;the set of measurement is analyzed, and the efficiency of new algorithm is validated by simulation.%为快速有效地求解介质粗糙面的反演问题,提出了一种能反演地表面参数的新方法--微多相粒子群算法.利用矩量法结合前后向迭代法快速求解双站散射系数.采用单纯形法实现有导向的约束初始种群的生成,以双站散射系数的测量值和理论计算值的偏差为目标函数,通过多相粒子群算法和单纯形相结合的算法对优化变量进行优化,使目标函数达到最小值实现地表参数的反演.给出了应用该方法的具体步骤,分析了测量结构,通过仿真实验验证了算法的有效性.