摘要
针对案例推理系统中案例检索的效率和质量问题,提出一种新的案例检索策略。采用粗糙集进行案例属性约简,完成案例库优化,并计算反映专家经验的属性权重,结合相似度计算和人工神经网络进行不同情况下的案例检索。运用UCI数据集进行了仿真对比,将其用于数字数据网故障诊断系统中,结果表明所提出的策略在不同数据集下均具有较高的检索效率,更加适用于实际CBR系统。
A new case retrieval strategy is proposed for case-based reasoning（CBR） system because of the case retrieval efficiency and quality problem. The rough set theory is adopted to implement case attribute reduction, complete the case base optimization, and compute attribute weights that reflect the expert＇s experience firstly, and then is combined with similarity computation and artificial neural network （ANN） to accomplish case retrieval in different situation. The UCI data set is used to simulate and compare. Application of the retrieval strategy in the data digital network fault diagnosis system indicates that the proposed case retrieval strategy has better performance in different data sets, and it is more fit for practical CBR system.
出处
《电讯技术》
北大核心
2010年第5期23-27,共5页
Telecommunication Engineering
关键词
基于案例推理
概率神经网络
粗糙集
案例检索
故障诊断系统
case-based reasoning（CBR）
probabilistic neural network
rough set
case retrieve
fault diagnosis system
