基于广义粗糙集与神经网络集成的旋转机械故障诊断研究

Fault Diagnosis of Rotating Machinery Based on Integration of Generalized Rough Sets and Neural Networks

故障诊断规则中判断条件的冗余、不完全和不确定性不利于实际应用。采用广义粗糙集理论对旋转机械振动故障诊断的非完备决策系统进行了约简 ,得到了更为简明的最优诊断规则 ;根据约简结果 ,建立了基于神经网络的故障诊断系统 ;网络的训练对比结果表明 ,基于粗糙集理论的约简处理简化了神经网络结构 ,提高了网络的训练效率 ;

In engineering applications, the incompleteness and redundancy in rules of fault diagnosis often lead to inconvenience. In this paper, rough sets theory was applied to reduction of incomplete diagnosis decision system of rotating machinery to find necessary conditions for diagnosis, and neural networks were used for fault pattern classification. Generalized rough sets theory and its application to reduction of incomplete decision system were introduced. Based on this theory, the incomplete fault diagnosis decision systems of rotating machinery were studied, and the optimal diagnosis rules were obtained. The application of the reduced diagnosis decision system to the neural fault classifier indicated that rough-sets-based-reduction reduces the dimension of input to neural network, and raises the efficiency of training. The practical examples validated the application of generalized rough sets integrated with neural networks to vibration fault diagnosis of rotating machinery.