基于混合粒子群算法和快速非均匀平面波算法的介质目标反演

提出了一种重构介质目标的新方法--混合粒子群算法，研究了几何形状已知的介质目标介电参数反演、均匀介质柱的外形轮廓反演及外形轮廓与介电参数均未知时的介质目标反演三类问题。利用快速非均匀平面波算法加速矩量法求解介质目标的雷达散射截面，以介质柱体的散射场的实际测量值与迭代计算值的偏差作为目标函数，通过单纯形法和伪群交叉算法混合的粒子群算法对优化变量进行优化，使目标函数达到最小值来对介质目标的介电特性进行电磁成像。仿真结果表明：混合粒子群算法简单、通用，在反演过程中不用加入正则化处理以确保数值稳定性，比简单遗传算法具有更好收敛性能、更高的成像精度和抗随机噪声干扰的能力。

Dielectric target reconstruction based on hybrid particle swarm optimization and the fast inhomogeneous plane wave algorithm

A novel approach for microwave imaging of the dielectric objects in free space using hybrid particle swarm optimization (HPSO) is presented. A scattering model based on the moment method(MOM) and the fast inhomogeneous plane wave algorithm （FIPWA） is applied to solve the scattering problem, and the inversions of three different objects are analyzed. The error between measured scattering data and computed scattering data is considered as the object function. The inverse scattering problem is transferred into an optimization problem by minimizing the object function with relative optimization parameters, which is solved by hybrid particle swarm optimization. Comparisons of the genetic algorithm (GA) and hybrid particle swarm optimization are carried out. The results show that hybrid particle swarm optimization is simple, versatile, and has the excellent performance in imaging precision and convergence. Another important advantage is that there is no necessary to utilize the regularization term which is essential to obtain the stabilization in application of typical direct optimization routine.