基于神经网络与模糊融合的外围故障诊断方法

针对用传统检测方法诊断模拟电路系统设备外围故障困难的问题,提出了一种利用BP神经网络与模糊融合相结合的故障诊断新方法,将神经网络与模糊融合结合起来,实现两者优势互补;首先利用神经网络的泛化能力对系统内部各可测点电压各用一个独立的BP神经网络对系统进行初级诊断,然后根据初级诊断结果,运用模糊融合诊断方法进行故障诊断,诊断结果更趋于合理,对模拟电路系统的外围故障实现正确定位;该方法能充分利用系统内部故障信息,有效避免采集外围设备信息的困难。

Outside Fault Diagnosis Based on Integrated Neural Network and Fuzzy Theory Fusion

In order to solve the difficulty of diagnosing outside faults in analog circuit system with traditional test method,a fault diagno- sis new approach for outside faults is developed based on neural network and fuzzy theory fusion using their superiority and avoiding their dis- advantages;First preliminary diagnosis for analog circuit system is performed separately by an independent BP network employing one acces- sible node test voltage data;Then according to the preliminary diagnosis results,fault diagnosis is accomplished by means of the proposed fuzzy data fusion method,the outcome of diagnosis is more reasonable and has the capability to realize correct fault location for outside faults in analog circuit system.The proposed approach can make full use of the fault information and avoid the difficulty of collecting peripheral e- quipment information.