论文部分内容阅读
针对转台的故障类型和特点,设计了基于邻域粗糙集的故障诊断系统。邻域粗糙集以邻域关系代替经典粗糙集的等价关系,通过邻域信息粒子逼近论域空间,可以直接处理数值型属性。提出了一种基于邻域等价关系的混合约简算法,通过计算条件属性的依赖度,对原始决策表进行约简,有效地剔除了系统的冗余属性。在约简决策表的基础提取分类规则,并最终生成分类机。通过对转台故障诊断的实验,验证了该方案的有效性和可行性。
Aiming at the fault types and characteristics of turntable, a fault diagnosis system based on neighborhood rough set is designed. The neighborhood rough set replaces the equivalence relation of the classical rough set by the neighborhood relation, and can approach the numerical properties directly by approximating the domain space by neighborhood information particles. A hybrid reduction algorithm based on neighborhood equivalence relation was proposed. By computing the dependency of conditional attributes, the original decision table was reduced and the redundant attributes of the system were effectively eliminated. The classification rules are extracted on the basis of the reduction decision table and finally the classifier is generated. Through the experiment of turntable fault diagnosis, the effectiveness and feasibility of the scheme are verified.