Modified Differential Evolution with Locality induced Genetic Operators for dynamic optimization

作者：Mukherjee Rohan; Debchoudhury Shantanab; Das Swagatam

来源：European Journal of Operational Research, 2016, 253(2): 337-355.

DOI：10.1016/j.ejor.2016.02.042

• Continuous optimization
• Dynamic optimization
• Differential Evolution
• Self adaptation
• Genetic operators

This article presents a modified version of the Differential Evolution (DE) algorithm for solving Dynamic Optimization Problems (DOPs) efficiently. The algorithm, referred as Modified DE with Locality induced Genetic Operators (MDE-LiGO) incorporates changes in the three basic stages of a standard DE framework. The mutation phase has been entrusted to a locality-induced operation that retains traits of Euclidean distance-based closest individuals around a potential solution. Diversity maintenance is further enhanced by inclusion of a local-best crossover operation that empowers the algorithm with an explorative ability without directional bias. An exhaustive dynamic detection technique has been introduced to effectively sense the changes in the landscape. An even distribution of solutions over different regions of the landscape calls for a solution retention technique that adapts this algorithm to dynamism by using the previously stored information in diverse search domains. MDE-LiGO has been compared with seven state-of-the-art evolutionary dynamic optimizers on a set of benchmarks known as the Generalized Dynamic Benchmark Generator (GDBG) used in competition on evolutionary computation in dynamic and uncertain environments held under the 2009 IEEE Congress on Evolutionary Computation (CEC). The experimental results clearly indicate that MDE-LiGO can outperform other algorithms for most of the tested DOP instances in a statistically meaningful way.

• 出版日期2016-9-1