基于模型诊断中的极小碰集问题是人工智能领域的一个重要课题，现实中很多实际问题都可以转化为极小碰集问题，如老师与课程问题，极小覆盖集问题等。通过对极小碰集问题特征的研究，本文结合粒子群优化求解极小碰集的算法提出了一个新的算法，来指导极小碰集的求解：引入学习机制，减少极小碰集求解中对无解空间的搜索；加入翻转策略，来加速极小碰集有解空间中的求解。实验结果表明本文提出的算法在求解极小碰集问题上的效率有明显提高。%In model-based diagnosis ,minimal hitting sets problem is an important topic of artificial intelligence ,and a lot of practical problems can be converted into it ,such as teachers and curricula problem ,minimum covering sets problem ,etc .By studying the characteristics of minimal hitting sets ,combining with the original PSO algorithm of computing minimal hitting sets ,this paper proposes a new algorithm to guide the minimum hitting sets computation :introducing a learning mechanism to cut down some search of the no-solution space;adding a flipping strategy to accelerate some solving of solution space .Experimental results show a signifi-cant improvement of this new algorithm in computing minimal hitting sets .
展开▼
