Memory Based Differential Evolution Algorithms for Dynamic Constrained Optimization Problems

A memory based differential evolution algorithm is adapted in this paper to solve Dynamic Constrained Optimization Problems. The approach is based on a mechanism to utilize the useful past information based on the problem special characteristics. A hybrid memory scheme combined short-term and long term memory is adopted based on this approach and reuses the best feasible individual and the best relaxed feasible individual found before changes. Moreover, an IATM method balancing the feasibility, a change detection mechanism based on trial vector, and the memory update operations are adapted to handling the changes on constraint or objective. Finally, the approach was tested on the recently proposed benchmark problems with 1000 change frequency. The results show that the proposed approach provides a very competitive performance with nine other state-of-the-art techniques.