改进多相粒子群算法反演粗糙面下方导体目标

提出了一种能快速有效反演介质粗糙面下方导体目标参数的方法--改进多相粒子群算法，优化以双站散射系数的测量值和理论计算值为偏差的目标函数，当目标函数达到最小值时，实现地下目标参数的反演。为加快反演过程，提高反演精度，正问题采用了快速互耦迭代算法这一数值算法来快速准确求解双站散射系数。逆问题采用了基于子群和母群交叉搜索机制以及小种群策略的改进多相粒子群算法，能减少目标函数计算次数（对应正问题计算次数）、提高全局寻优能力。文中反演了一维粗糙面下方截面为圆柱和任意连续形状的导体目标，仿真实验验证了算法具有较好的反演精度和较强的抗随机噪声能力。

Inversion of PEC targets below dielectric rough surface based on hybrid multi-phase particle swarm optimization

A hybrid multi-phase particle swarm optimization (HMPPSO) is proposed to the inversion of perfect electric conductor (PEC) targets below the dielectric rough surface. While the errors between the measured bistatic scattering coefficients and the computed bistatic scattering coefficients are considered as the object function, the parameters of the buried object are viewed as the variables to be optimized. The buried object is inversed when the object function is minimized to the minimal. To quicken the inversion procedure and improve the inversion accuracy, a fast cross coupling iterative approach (CCIA) is used to solve the forward scattering problem and the HMPPSO, which is based on a small swarm size strategy and cross searching mechanism among the sub-swarms and optimal swarm, is used to reduce the object function evaluation times (corresponding to the forward problem calculation times) and improve the global searching ability. The PEC objects with the cross section of circular and of any irregular continuous shape below one dimensional rough surface are inversed and both the accuracy and the robust of anti-noise of the algorithm are validated by simulation.