In view of Elman neural network for bearing fault diagnosis constrained by a significant amount of redundant information and the subjectivity of attribute discretization method based on information entropy in threshold selection,an improved discretization algorithm for continuous attribute in rough set based on information entropy is proposed.The al-gorithm combined with rough set theory is used for bearing failure data reduction,excluding the attribution samples that have little impact on the decision-making and building bearing fault diagnosis system based on optimized rough set-Elman neural network.The method is applied to the bearing fault diagnosis and the results show that the improved El-man network,while ensuring the diagnostic accuracy,has better efficiency in the diagnosis.%针对Elman神经网络在轴承故障诊断中受到大量冗余信息制约的问题，以及基于信息熵的属性离散方法阈值选取具有主观性的缺点，提出了改进的基于信息熵的粗糙集连续属性离散化算法。利用该算法结合粗糙集理论对轴承故障数据进行约简，剔除对决策影响较小的属性样本，构建优化的粗糙集-Elman神经网络轴承故障诊断系统。将该方法应用到轴承故障诊断中，结果表明，改进后的Elman网络在保证诊断准确性的同时具有明显的效率优势。
展开▼
