概率神经网络在导弹自动驾驶仪故障诊断中的应用

Application of Probabilistic Neural Network in Autopilot System Fault Diagnosis

首先分析基于PNN网络进行某型导弹自动驾驶仪故障诊断的可行性,并研究诊断方法的思路和建立故障诊断网络的过程。然后建立自动驾驶仪I通道故障诊断的PNN网络,并进行诊断仿真。仿真结果表明,建立的故障诊断PNN网络具有较高的故障诊断正确率,诊断效果要优于BP神经网络,满足用于导弹自动驾驶仪故障诊断的要求。

The feasibility of fault diagnosis to channel I circuit of autopilot system using Probabilistic Neural Network method is analysed. Then the novel method routine and classifies faults are proposed. Simulation and com- pare of its performance show this network based on PNN method bears higher diagnosis accuracy compared to BP neural network and can meet the need of application in autopilot fault diagnosis.