算法的优化:基于适应度方差的涡流搜索方法

A fitness variance based vortex search method for algorithm optimization

研究了算法的优化方法。为克服涡流搜索(VS)方法在优化运行时可能会陷入局部最优的问题,提出了一种基于适应度方差(FV)的涡流搜索(VS)方法,简写为FVS算法。该方法每次迭代时计算一次全体备选解适应度值的方差大小,如果适应度方差小于某一设定阈值,则调整涡流搜索方法备选解产生的位置,使方法产生的备选解能够跳出局部最优的限制。将这种基于适应度方差的涡流搜索方法与原始涡流搜索方法和其他几种优化算法对8个测试函数进行了寻优对比,结果验证了这种FVS方法的有效性。

The algorithm optimization is studied. In order to overcome the problem that the vortex search（ VS） method for algorithm optimization may fall into local optimum in operation,a new vortex search method based on fitness variance,called the FVS,is proposed. The FVS method calculates the variance of fitness value of all alternative solutions at each iteration. If the calculation result is less than a set threshold,the position of the alternative solution of the vortex search method is adjusted. The alternative solution generated by the method can jump out of the limit of the local optimal. This optimization method is compared with the original vortex search method and other methods on 8 test functions,and the results verify the effectiveness of the proposed method.