Abstract: There exist some problems that the traditional Artificial Bee Colony algorithm（ABC） and its extension in multi object（MOABC） has a slow convergence speed, easily falling into local minima, optimization accuracy lost and other issues under the condition of high dimension, multi peak function. Based on the characteristic of knee points that it can improve convergence and distribution, an algorithm that rapidly identificates the knee points is designed in this paper and applied to the MOABC, it proposes the improved Multi-Objective Artificial Bee Colony algorithm based on the strategy of Knee point（KnMOABC）. In the iterative process, the pareto dominating relation is taken into account firstly, and the knee points are selected as the individuals for next generation, which greatly enhances the convergence speed of the algorithm, at the same time, an adaptive strategy is added into the knee point recognition algorithm to ensure the distributivity of the algorithm. The experimental results show that the performance of KnMOABC is better than that of the three latest multi-objective artificial bee colony algorithm.

Key words: multi-objective artificial bee colony algorithm, high-dimensional and multimodal functions, knee points, adaptive identification strategy

摘要： 传统的人工蜂群算法（Artificial Bee Colony algorithm，ABC）及其在多目标上的扩展（Multi Objective Artificial Bee Colony algorithm，MOABC）存在着在高维、多峰函数情况下收敛速度变慢、后期容易陷入局部最优以及寻优精度丢失等问题。基于knee points提高收敛性和分布性的特点，设计了一种快速识别knee point的算法并将其应用到多目标人工蜂群算法中，提出了一种基于knee points的改进多目标人工蜂群算法（KnMOABC）。算法在迭代过程中考虑pareto支配关系的同时，优先选择knee point作为下一代个体，极大地增强了算法的收敛速度，同时，在knee point识别算法中加入自适应的策略以保持良好的分布性。实验结果表明，KnMOABC的性能优于三个最新的多目标人工蜂群对比算法。

关键词: 多目标人工蜂群算法, 高维多峰函数, knee points, 自适应识别策略