量子扩展蚁群连续优化改进算法

Improved quantum extended ant colony algorithm for continuous optimization

针对扩展蚁群算法收敛精度不高、容易陷入局部最优和出现收敛停滞等缺点,提出量子扩展蚁群连续优化改进算法。分析扩展蚁群算法可行解的更新与产生机制;在此基础上,引入量子比特作为蚂蚁位置信息的载体,增加解的多样性;采用云模型自适应产生高斯核函数采样的标准差,优化高斯采样结果,加速优化进程和最优解的搜索;根据优化进程自适应调整采样函数的选择概率,丰富采样的样本;结合云模型控制的变异策略及量子非门等局部寻优手段,有效避免种群早熟。

Aiming at disadvantages of extended ant colony optimization algorithm including low convergence accuracy, easily falling into local optimum and convergence stagnation phenomenon, an improved quantum extended ant colony optimization algorithm for continuous optimization was proposed. The mechanisms of updating and generating for feasible solutions in ACO~ algo- rithrn were analyzed. On this basis, the quantum bits were introduced as ants location information carrier, the diversities of solutions were increased. Using cloud model adaptively to generate standard deviation of sampling of the Gaussian kernel function, the Gaussian sampling results were optimized, and the whole optimization process and the optimal solution searching process were speeded up. Sampling function selection probabilities were adaptively adjusted according to the optimization process, sampiing samples were enriches. Local optimization methods were integrated, such as mutation strategies controlled by cloud model, quantum NOT gate and quantum crossover, the premature of population was avoided effectively.