作　　者： (曹如胜）; (倪世宏）; (张鹏）;

机构地区： 空军工程大学航空航天工程学院,西安710038

出　　处： 《计算机科学》 2017年第9期239-242,共4页

摘　　要： 针对贝叶斯网络结构学习对算法高效性的要求,提出将云遗传算法和模拟退火算法相结合的云遗传模拟退火算法,以云遗传算法的选择、云交叉和云变异来完成模拟退火算法中的更新解操作;同时,针对算法在特定条件下陷入早熟收敛的问题,提出了改进的云交叉算子和云变异算子。仿真实验结果表明,所提云遗传模拟退火算法能有效提高贝叶斯网络学习的效率和准确性。 In view of the highly active requirement of Bayesian networks structure learning, a learning strategy was pro- posed based on cloud genetic annealing algorithm which combines cloud genetic algorithm and simulated annealing algo- rithm. Update solution operation are accomplished by selection,cloud cross and cloud variation. In view of the shortco- mings of algorithm being involved into the local optimization untimely, this paper put forward an adaptive cloud cross- over operation and cloud mutation operator. The simulation shows that the accuracy of learning and operational efficien- cy are increased.

关 键 词： 云模型 遗传算法 模拟退火 结构学习