概念格理论属性约简算法研究

摘要

Concept lattice is a concept hierarchical structure,which is established according to binary relationship between data objects and attributes. It vividly and simply reflects the generalization and specialization relationship between concepts. This feature of concept lattice is used for attribute reduction of fault diagnosis. The reduction results are transmitted to neural network to diagnose the faults. The diagnosis results show that the existing fault types can be identified correctly using the core attributes and relatively necessary attributes and reduce the dimension of fault diagnosis parameters,which accelerates the algorithm's operating speed.%概念格是根据数据集中对象与属性之间的二元关系建立的一种概念层次结构,生动简洁地体现了概念之问的泛化和特化关系.利用概念格的这一特性,提出利用概念格理论进行故障诊断时的属性约简.为验证属性约简的正确性,将约简结果送到神经网络中进行故障诊断,诊断结果表明:利用概念格理论所得到的核心属性和相对必要属性可以对现有故障类型进行正确辨识,降低了故障诊断参数的维数,有利于加快诊断算法的运算速度.