Abstract: For the problem of general genetic algorithm in solving Assembly Sequence Planning (ASP) problem with slow search speed and massive repeated solutions,a new method for ASP was proposed in which ant colony algorithm,chaos algorithm and genetic algorithm were mixed,and the strategy of dynamic update population number was also introduced.The information matrix of evaluation index was established by using the information between the parts based on disassembly interference matrix.In earlier stage,ant colony algorithm was used for local search,and chaos-genetic algorithm was used for global search in the later stage.In the global search process,Logistic mapping was used to generate offspring of assembly sequence to avoid generating repeated solutions,and a certain proportion was set to select the searched solutions and update population number dynamically.To further improve the quality of discrimination sequence,a new evaluation index named coherence was introduced in which the application times of assembly tools and the changing times of assembly direction were combined,and aid tools were also introduced at the same time.The gear pump was taken as an example to compare with ant colony algorithm,genetic algorithm and chaos algorithm,and the proposed algorithm was more effective through analysis of the performance from aspects of fitness value,running time and convergence speed.

Key words: assembly sequence planning, ant colony algorithm, genetic algorithms, chaos algorithm, information matrix

摘要： 针对遗传算法在求解装配序列规划问题中速度慢、产生重复解等问题,提出一种将蚂蚁算法、混沌算法和遗传算法结合,加入动态更新种群数目这一策略的新方法。利用拆卸干涉矩阵获取零件之间的信息,建立评优指标的信息矩阵。算法前期采用蚂蚁算法进行局部搜索,后期运用混沌—遗传算法进行全局搜索。在全局搜索中,采用Logistic映射生成子代装配序列以避免产生重复解,并设置一定的比例筛选搜索结果,动态更新种群数目。为进一步提高序列优劣的区分度,将装配工具的运用次数与装配方向的改变次数合并为连贯性,同时引入辅助工具这一指标。以齿轮油泵为例,将所提算法与蚂蚁算法、遗传算法和混合算法进行比较,从适应度值、运行时间和收敛速度方面进行分析,验证了该算法的有效性。

关键词: 装配序列规划, 蚂蚁算法, 遗传算法, 混沌算法, 信息矩阵