基于概率神经网络的潜舰武器系统故障诊断

Application of intelligent detection in submarine weapon system fault diagnosis

针对传统故障诊断方法检测某型潜舰导弹武器系统故障准确率不高、耗时长的问题,提出基于概率神经网络的智能诊断方法。介绍该网络的典型结构及优势所在,以某型潜舰导弹武器系统为验证对象,选取合适特征向量、归纳合理故障类型、建立相应神经网络,并运用Matlab仿真验证。结果表明在现有数据库中,概率神经网络对该系统的故障诊断正确率为77.8%。这表明基于概率神经网络的故障诊断基本能够区分该系统故障类型,大大减少了部队故障诊断时间和人力投入。

Aiming at the problems of low accuracy and long time for traditional fault diagnosis methods to detect the fault of a submarine missile weapon system,an intelligent diagnosis method based on probabilistic neural network was pro-posed.This paper introduces the typical structure and advantages of the network,takes a certain submarine missile weapon system as the verification object,selects the appropriate feature vector,summarizes the reasonable fault types,establishes the corresponding neural network,and uses Matlab simulation to verify.The results show that:in the existing database,the probab-ilistic neural network fault diagnosis accuracy of the system is 77.8%.This shows that the fault diagnosis based on probabil-istic neural network can effectively distinguish the fault types of the system,and greatly reduce time and manpower input of the armyfault diagnosis.