随动系统模拟电路模块化故障诊断研究

针对人工神经网络在对装备电路进行故障诊断时出现的网络结构庞大，而影响网络诊断速度的问题，提出了模块化诊断思想。首先根据电路的功能特点以及系统电路分割的原则，将装备电路分割成具有独立功能的模块及子模块，然后建立起混联、串联2种模型，分别对其提出逻辑思想、模糊信息融合的诊断方法，从而将电路故障快速、有效地定位在模块级别，如需进一步诊断元器件故障，只需对故障子模块建立规模较小的诊断网络，从而有效避免诊断大规模电路时诊断网络结构庞大的问题。

Modular Fault Diagnosis for Servo System Analog Circuit

When network structure ward. First by the may be divided in artificial neural network is used to fault diagnosis for large-scale analog circuit, the huge is bad for the speed of fault diagnosis, so the modular diagnosis method is brought forfunction character of circuit and the principle of circuit partition, the equipment circuit to some module circuits or child-modules which has independent function; then it establishes two kind of circuit model such as series connection, hybrid connection for module-level fault, and separately logical thinking will be adopted, fuzzy information fusion to diagnose the module-level fault fleetly and effectively, if it is necessary of diagnosis component-level fault, little-scale diagnosis network is set up only for fault child-module circuit, so the question of huge diagnosis network structure is avoided effectively for diagnosing large-scale analog circuit.