基于粗糙集的多层容错神经网络故障诊断

Multilayer Fault-tolerance Neural Networks Fault Diagnosis Based on Rough Sets

针对神经网络故障诊断问题中输入属性维数多和数据量庞大的情况,首先利用粗糙集理论对原始数据进行约简,并按照一定的原则选取多个约简;然后对所得到的多个约简分别构建子神经网络,将多个子网络合成统一的容错网络。结合实例应用取得了令人满意的结果,并为高可靠性设备的故障诊断提供了新的思路。

To the condition of many input dimensions and lots of data in neural networks fault diagnosis, some reductions from data based on rough sets theory are derived and more than one reduction according to some criteria are choosed; then subnets by the reductions are built, and then a general fault-tolerance neural networks by synthesizing each subnet is built. A fault tree in the process of fault diagnosis is used. Finally, an application is realized by this method, gratifying result is achieved and a new idea is provided to diagnosis of high reliability equipments.