MOEA/D with Adaptive IWO for Synthesizing Phase-Only ReconfigurableLinear Arrays

摘要

In order to well solve the phase-only reconfigurable arrays synthesis problems, we introduce an adaptive strategyin invasive weed optimization (IWO), and integrate the adaptive IWO (AIWO) into the framework of MOEA/D, apopular multi-objective algorithm. Then, a new version of MOEA/D with adaptive IWO, named MOEA/D-AIWO is proposedin this paper for solving the synthesis problems. In MOEA/D-AIWO, the proposed adaptive strategy is adopted forimproving search ability and balancing diversity and convergence. We introduce an adaptive standard deviation, whichchanges not only with the increase of evolution generations, but also exponentially with the fitness function value of eachindividual. This strategy improves the convergence rate and helps the seeds escape from local optimum. Taking advantageof the powerful searching ability of invasive weeds and well framework of MOEA/D, the overall performance of the proposedMOEA/D-AIWO is illustrated in solving two sets of phase-only reconfigurable arrays synthesis problems. Comparingresults with MOEA/D-IWO (MOEA/D with original IWO) and MOEA/D-DE are also provided in this paper.