Abstract: A discrete particle swarm algorithm based on value ordering of min-conflict heuristic is proposed. It uses the min-conflict heuristic to select values from domains instead of random selection. This strategy searches the promising solution space for global solution when the particles exploit the search space. It tests the hybrid algorithm with random constraint satisfaction problems. The experimental results show that the hybrid particle swarm algorithm can converge the global solution faster and improve the performance either on the iterations or runtime several-fold.

Key words: Particle swarm, Binary constraint satisfaction problem, Mini mizing conflicts heuristic, Value order

摘要： 提出了一个基于最小冲突启发式值序的二元约束满足问题粒子群算法，利用值序对值的选取方式代替随机选择的盲目搜索方式，使群体在探索解空间的时候，选择有希望能找到全局解的地方搜索。使用随机约束满足问题的实验表明，改进后的算法比原算法能以更快的速度收敛到全局解，无论在迭代次数还是运行时间上均能数倍提高算法的效率。

关键词: 粒子群算法, 二元约束满足问题, 最小冲突启发式, 值序

CLC Number: 

• TP311