A New Multi-region Modified Wind Driven Optimization Algorithm with Collision Avoidance for Dynamic Environments

摘要

This paper describes a new approach to deal with dynamic optimization that uses a multi-population. Its main features include the use of a modified wind driven optimization algorithm that aims to foster impact of pressure on velocities of particles. Moreover, a concept of multi-region inspired from meteorology has been introduced along with a new collision avoidance technique to maintain good diversity while preventing collision between sub-populations. The method has been assessed using Moving Peaks Benchmark and compared to state of the art methods. Preliminary results are very encouraging and show viability of the method.