论文部分内容阅读
针对故障诊断信息的不一致性,提出一种基于粗糙集决策网络的故障规则提取方法。将故障诊断决策系统通过分辨矩阵和分辨函数进行属性约简后,构造出一个不同简化层次的决策网络。将属性约简集作为网络初始节点,根据网络节点得到决策规则集;同时,为了有效滤除噪声,在置信度的基础上引入了规则覆盖度的概念,对提取的规则进一步评价,最终提取有效的诊断规则。旋转机械故障实例验证了该方法的有效性。
Aiming at the inconsistency of fault diagnosis information, a fault rule extraction method based on rough set decision network is proposed. After the fault diagnosis decision system performs attribute reduction through the resolution matrix and the resolution function, a decision network of different simplified levels is constructed. At the same time, in order to effectively filter out the noise, the concept of rule coverage is introduced on the basis of confidence level to further evaluate the extracted rules and finally extract the effective rules The diagnostic rules. The example of rotating machinery failure validates the effectiveness of this method.