Efficient Multiobjective Optimization of Time-Modulated Array Using a Hybrid Particle Swarm Algorithm With Convex Programming

An efficient hybrid method for multiobjective optimization (MO) of time-modulated array (TMA) is proposed. First, by making best use of the convexity in the TMA optimization problem, a set of reference points for Pareto front (PF) are generated by convex programming. Then, with the obtained reference points function as the prior knowledge for the initial strategy of multiobjective particle swarm optimization (MOPSO), the search of the MOPSO is located around the most interested region while the diversity of the population is also guaranteed to achieve a wide spread of PF. With the proposed hybrid method, the multiobjective optimization (MO) of relative large-scale TMA (comparing with other existing TMA MO method) can be achieved in a limited computational budget while the quality of the obtained PF is much better than the traditional nonhybrid method, thanks to the appropriate use of the problem prior knowledge. Examples with 50- and 100-elements linear TMA are presented to demonstrate the robustness of the proposed method to deal with the relative large scale multiobjective TMA optimization problems.