A novel differential evolution scheme combined with particle swarm intelligence
Xing Song Xu, Yuanxiang Li, Shenlin Fang, Yu Wu, Feng Wang · 2008
Differential evolution (DE) and particle swarm optimization (PSO) are the evolutionary computation paradigms, and both have shown superior performance on complex nonlinear function optimization problems. This paper detects the underlying relationship between them and then qualitatively proves that the two heuristic approaches from different theoretical background are consistent in form. Within the general perspective, the PSO can be regarded as a kind of DE. Inspired by this, a novel variant of DE mixed with particle swarm intelligence (DE-SI) is presented. Comparison experiments involving ten test functions well studied in the evolutionary optimization literature are used to highlight some performance differences between the DE-SI, two versions of DE and two PSO variants. The results from our study show that DE-SI keeps the most rapid convergence rate of all techniques and obtains the global optima for most benchmark problems.