模糊神经网络及其在控制系统故障诊断中的应用

Fuzzy-Neural Network and Its Application to Fault Diagnosis of the Control System

针对控制系统中模拟电路故障诊断时的不确定性问题,提出了将模糊理论和神经网络相结合的方法;首先利用模糊理论描述不确定性信息,然后利用人工神经网络完成不确定性推理,最后利用模糊理论对推理结果进行解释和决策,从而得出故障诊断结论;结合某型船舶主机遥控系统中延时电路板的故障诊断问题,阐述了将该方法应用于实际控制系统的故障诊断过程;结果表明,该方法能够较好地处理模拟电路故障诊断过程中的不确定性问题,有效地提高故障模式的识别能力,将故障准确地定位到元器件。

In order to handle the uncertainties in fault diagnosis of analog circuits used in the control system, a method, which combines fuzzy theory and neural network, is proposed. First, fuzzy theory was applied to describe the uncertainty information. Then, uncertainty reasoning was accomplished by using artificial neural network. At last, the reasoning results were explained and decision was made by using fuzzy theory. So, the results of fault diagnosis can be obtained. Combined with the fault diagnosis system of time delay board, which is used in the remote--control system of submarine＇s main marine engine, the process to apply this method to fault diagnosis of the control system was described in detail. The results show that this method can properly dispose the uncertainties in fault diagnosis of analog circuits, improve the recognition ability of fault patterns and locate the fault component.