一种基于SOFM网络的模拟电路模糊故障定位方法

针对传统的模拟电路故障诊断方法的局限性,本文研究了一种基于自组织特征映射(SOFM)神经网络的模糊故障诊断方法。该方法能够将SOFM神经网络竞争的特点与模糊推理系统的自组织和自学习优点相结合。将模拟电路分成易于诊断的子网络,将模拟电路的模糊故障诊断系统与SOFM神经网络和模糊推理系统相结合,利用故障诊断系统快速定位故障节点。研究表明,该方法能实现快速定位模拟电路故障,拥有较高的实用价值。

Aiming at solving the limitations of the traditional analog circuit fault diagnosis methods, this paper studies a fuzzy locating method for analog circuits faults, which is based on SOFM neural network method. Integrating the competitive characteristic of SOFM neural network with the fuzzy inference system advantages of self-organizing function and self-learning function, this method is able to tear large analog circuits into sub networks which are easy to he diagnosed. And this method can make use of the analog circuit fault fuzzy diagnosis system, which is a combination of SOFM neural network and fuzzy inference system. So, it can locate the fault subnetwork and find out the fault node in a high speed. It is proved by examples that this method can locate the fault of analog circuit quickly and has great practical value.