中文摘要：

采用传统故障诊断方法对复杂模拟电路进行故障诊断，存在诊断速度慢、故障点难以准确定位等缺点，为此提出基于神经网络理论的故障状态分类器设计方法，以某型飞机自动驾驶仪854C放大器为例，对故障状态分类器进行了分析设计和仿真研究。仿真结果显示：所设计的基于神经网络理论的故障状态分类器实现了对此类航空设备模拟电路故障的快速、准确和有效的识别与定位。

英文摘要：

In view of the traditional fault diagnosis methods had the shortcomings that the diagnosis speed was slow and it was hard to accurately fix the faults when it was diagnosing the complicated analog circuit, the fault-state classing implement design method based on the artificial neural network theory were introduced here. Then, with some mould aircraft autopilot＇s 854C amplifier as the exampie, the design method and the real diagnosis course was analyzed and explained, and so on, the simulation research was made. The high-speed diagnosis is realized. The results show that this faultstate classing implement is effective to solve the issue.