SDG故障诊断规则的知识约简  

Fault Diagnosis Rule Based on SDG and Knowledge Reduction

作　　者：谢刚[1] 赵静阁[1] 詹峰[1] 谢克明[1] 

机构地区：[1]太原理工大学信息工程学院,太原030024

出　　处：《小型微型计算机系统》2011年第9期1881-1884,共4页Journal of Chinese Computer Systems

基　　金：国家自然科学基金项目(60975032)资助;山西省青年基金项目(2010021016-1)资助

摘　　要：符号有向图(SDG)是揭示流程系统深层知识的定性模型,用于描述流程系统的状态变量及其变量间的故障信息传递关系.当系统的状态变量过多,运用SDG故障诊断算法生成的故障规则过于庞大,推理困难.粒矩阵的知识约简算法能有效约简冗余属性.因此,将粒矩阵的知识约简算法引入SDG故障诊断,以电站除氧器系统为例,使用粒矩阵的知识约简算法约简主要故障的故障规则,简化规则中的冗余节点,提高故障诊断效率,最后验证了约简后的故障诊断规则的正确和有效.Signed directed graph （ SDG ） is an important qualitative model that can be used to express the deep knowledge of the process industry, and describe the state variables and their cause-effect relations in the system. However, fault diagnosis rules using SDG fault diagnosis algorithm has enormous redundancy because of too many system state variables, which can lead to the fault reasoning difficultly. Redundant attribute can be reduced by knowledge discovery algorithm of granular computing effectively. Consequently, SDG-based fault diagnosis combines with granular computing is proposed in this paper. Then the power plant deaerator is taken for example, and the fault rules of main fault are reduced by the method which improves the fault diagnosis effectively. Finally, the fault diagnosis conclusion illustrates this method concisely and suitably.

关 键 词：符号有向图 故障诊断 粒矩阵 知识约简 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]