基于差分进化的回溯搜索优化算法研究与改进

Study and improvement of backtracking search optimization algorithm based on differential evolution

针对回溯搜索优化算法收敛速度慢和易早熟的缺点,提出了一种改进算法。首先,利用麦克斯韦分布产生变异尺度系数,并在此基础上提出了一种新的变异算子。新变异算子有效地加快了收敛速度。同时,在变异策略中添加了一种选择机制以增加全局搜索能力,避免出现早熟收敛。通过与差分进化的变异策略对比和经典测试函数的测试,实验结果表明改进算法不仅具有较快的收敛速度,而且具有良好的全局搜索能力。

For slow convergence and easiness to trap in local optimum of backtracking search optimization algorithm, this paper presented an improved algorithm. It used Maxwell-Bohzmann distribution to generate mutation scale factor and redesigned the mutation strategy based on Maxwell-Bohzmann distribution, which improved convergence speed effectively. Moreover, it added a selection mechanism to mutation strategy to enhance global search ability, which could avoid premature convergence. Through comparing with mutation strategy of differential evolution and numerical experiments on a suite of standard test functions, the fact shows that the improved algorithm not only has faster convergence speed, but also good global search capability.