基于粗糙集的容错神经网络故障诊断系统

Tolerance neural net fault diagnosis system based on rough set

粗糙集和神经网络在故障诊断中都得到了广泛的应用,但两者都有其局限性,同时在许多方面有其互补性,融合粗糙集和神经网络各自的优势,建立了粗糙集——容错神经网络故障诊断系统。利用粗糙集对原始数据进行简约,导出最简诊断规则,根据选择的冗余约简和最简诊断规则建立粗糙集——容错神经网络故障诊断系统。以滚动轴承故障诊断为例,仿真结果表明系统提高了故障诊断准确率和诊断速度,消除了故障诊断中的误报和漏报现象。

Rough set and neural network once plays an important role in fault diagnosis and so on, but there exist some limits when they are simply applied. On the other hand, because they are complementary, the superiority of rough set and neural net were amalgamed, and rough set-tolerance neural net fault diagnosis system was put forward. The reductions from crude date on rough sets theory was driven, and the minimal diagnosis rule was gat, then rough set-tolerance neural net fault diagnosis system was built. This system to the fault diagnosis of rolling bearings was applied. Simulation results indicate that the system has increased the quality and rate of diagnosis, andcan eliminate the effects of misinformation and failing to report on the quality of diagnosis in fault diagnosis.