融合佳点集机制的动态搜索烟花爆炸搜索算法

摘要

For overcoming the weakness of firework explosion search ( FES) algorithm in being prone to prematurity and improving its solu-tion performance, we proposed in this paper a dynamic firework explosion search algorithm which integrates the mutation mechanism of good-point set.First, in order to improve the precision of solution, every iteration process was targeted at current best individual to execute dynam-ic random search and this enhanced the local search for the current best.On the other hand, when the overcrowding degree of population ex-ceeded the preset thresholdλ, all the individuals, except 10%excellent ones remained, were to reinitialise based on good-point set mecha-nism to help the population get rid of the constraint from local optimum.In the end, the experiments on six classical Benchmark functions demonstrated that the improved FES algorithm could fast converge, prevented the prematurity and had better robustness.%为了克服烟花爆炸搜索算法容易早熟的弱点，提高其求解性能，提出一种融合佳点集变异机制的动态搜索烟花爆炸算法。首先为了提高算法的求解精度，每一次迭代过程均针对当前最佳个体执行动态随机搜索，加强对当前最佳的局部搜索。另一方面，当种群的拥挤程度超越设定的阈值λ时，除保留10％的优秀个体外，其余个体基于佳点集机制进行重新初始化，帮助种群摆脱局部最优的约束。最后，在6个Benchmark函数上的实验表明，该算法能快速收敛、克服早熟，并且具有较佳的鲁棒性。