针对人工蜂群算法在求解过程中存在收敛速度慢、易陷入局部最优解等缺点,提出了基于加强局部搜索策略的人工蜂群算法(ABC Based On Enhancing Local Search Ability,LSABC).一方面,在雇佣蜂搜索阶段,利用两种不同的搜索公式得到两组解,...针对人工蜂群算法在求解过程中存在收敛速度慢、易陷入局部最优解等缺点,提出了基于加强局部搜索策略的人工蜂群算法(ABC Based On Enhancing Local Search Ability,LSABC).一方面,在雇佣蜂搜索阶段,利用两种不同的搜索公式得到两组解,并将适应度最佳者作为候选解,增加解的多样性;同时,在搜索公式中加入个体的双重认知能力平衡算法的勘探和开发能力.另一方面,在侦察蜂搜索阶段,采用禁忌搜索策略,将局部极值存入禁忌表中,帮助算法跳脱局部最优解,达到避免算法早熟的同时加快算法收敛速度的目的.由于LSABC算法的改进与粒子群算法相似,为验证LSABC算法的寻优性能,针对8个经典基准函数,选取标准ABC算法、PSO算法、EABC算法、RLPSO算法及LSABC算法分别进行对比测试.计算实验结果表明,LSABC算法在求解精度和收敛速度方面明显提高,易于跳脱局部最优解.展开更多
Computer-aided process planning （CAPP） is an essential component of computer integrated manufacturing （CIM） system. A good process plan can be obtained by optimizing two elements, namely, operation sequence and th...Computer-aided process planning （CAPP） is an essential component of computer integrated manufacturing （CIM） system. A good process plan can be obtained by optimizing two elements, namely, operation sequence and the machining parameters of machine, tool and tool access direction （TAD） for each operation. This paper proposes a novel optimization strategy for process planning that considers different dimensions of the problem in parallel. A multi-dimensional tabu search （MDTS） algo-rithm based on this strategy is developed to optimize the four dimensions of a process plan, namely, operation sequence （OperSeq）, machine sequence （MacSeq）, tool sequence （TooISeq） and tool approach direction sequence （TADSeq）, sequentially and iteratively. In order to improve its efficiency and stability, tabu search, which is incorporated into the proposed MDTS al- gorithm, is used to optimize each component of a process plan, and some neighbourhood strategies for different components are presented for this tabu search algorithm. The proposed MDTS algorithm is employed to test four parts with different numbers of operations taken from the literature and compared with the existing algorithms like genetic algorithm （GA）, simulated annealing （SA）, tabu search （TS） and particle swarm optimization （PSO）. Experimental results show that the developed algo-rithm outperforms these algorithms in terms of solution quality and efficiency.展开更多
文摘针对人工蜂群算法在求解过程中存在收敛速度慢、易陷入局部最优解等缺点,提出了基于加强局部搜索策略的人工蜂群算法(ABC Based On Enhancing Local Search Ability,LSABC).一方面,在雇佣蜂搜索阶段,利用两种不同的搜索公式得到两组解,并将适应度最佳者作为候选解,增加解的多样性;同时,在搜索公式中加入个体的双重认知能力平衡算法的勘探和开发能力.另一方面,在侦察蜂搜索阶段,采用禁忌搜索策略,将局部极值存入禁忌表中,帮助算法跳脱局部最优解,达到避免算法早熟的同时加快算法收敛速度的目的.由于LSABC算法的改进与粒子群算法相似,为验证LSABC算法的寻优性能,针对8个经典基准函数,选取标准ABC算法、PSO算法、EABC算法、RLPSO算法及LSABC算法分别进行对比测试.计算实验结果表明,LSABC算法在求解精度和收敛速度方面明显提高,易于跳脱局部最优解.
基金supported by the State Key Program of National Natural Science Foundation of China (Grant No. 51035001)National Natural Science Foundation of China (Grant Nos. 50825503, 50875101)
文摘Computer-aided process planning （CAPP） is an essential component of computer integrated manufacturing （CIM） system. A good process plan can be obtained by optimizing two elements, namely, operation sequence and the machining parameters of machine, tool and tool access direction （TAD） for each operation. This paper proposes a novel optimization strategy for process planning that considers different dimensions of the problem in parallel. A multi-dimensional tabu search （MDTS） algo-rithm based on this strategy is developed to optimize the four dimensions of a process plan, namely, operation sequence （OperSeq）, machine sequence （MacSeq）, tool sequence （TooISeq） and tool approach direction sequence （TADSeq）, sequentially and iteratively. In order to improve its efficiency and stability, tabu search, which is incorporated into the proposed MDTS al- gorithm, is used to optimize each component of a process plan, and some neighbourhood strategies for different components are presented for this tabu search algorithm. The proposed MDTS algorithm is employed to test four parts with different numbers of operations taken from the literature and compared with the existing algorithms like genetic algorithm （GA）, simulated annealing （SA）, tabu search （TS） and particle swarm optimization （PSO）. Experimental results show that the developed algo-rithm outperforms these algorithms in terms of solution quality and efficiency.
