双尺度协同变异的离散粒子群算法

Discrete particle swarm optimization based on double-scale cooperation mutation

针对离散粒子群算法局部搜索能力差的不足,提出了一种基于双尺度协同变异的离散微粒群算法.采用对当前最优解进行双尺度速度变异的方法,来实现提高算法局部最优解搜索和改善最优解精度的能力.在算法初期利用大尺度速度变异可增加粒子的多样性,快速定位到最优解区域;算法后期则通过逐渐减小的小尺度变异可提高算法在最优解附近的局部精确解搜索性能.将改进算法应用于5个标准Benchmark函数优化问题,并与其他5种离散粒子群算法在收敛速度和稳定性方面进行比较,统计结果表明新算法具有更加的优化性能.

To deal with the problem in discrete particle swarm optimization of the particles searching blindly and not being able to carry out a deep local search around the current optimal solution, a discrete particle swarm optimiza- tion （DPSO） algorithm based on double-scale cooperation velocity mutation was proposed. The double-scale velocity mu- tation operator was introduced for the current optimal solution, which can not only improve the local search function, but also increase the precision of the optima solution. The coarse-scale mutation operator can be utilized to quickly localize the global optimized space at early evolution. The novel scale-changing strategy produced a smaller fine-scale mutation operator according to the evolution and developed mutation operators with fine-scale possibilities to implement a local ac- curate minima solution search at the late evolution stage. The experimental studies on five standard benchmark functions and the experimental results show that the proposed method can not only effectively solve the p^oblem of a lack of local search ability, but also significantly speed up the convergence while improving the stability.