基于蚁群神经网络的发射系统故障诊断  

Research on Fault Diagnosis for Launch System Based on Ant Colony Neural Network

作　　者：宋涛[1] 舒涛[1] 雷荣强[1] 刘赞[1] 

机构地区：[1]空军工程大学防空反导学院,西安710051

出　　处：《火力与指挥控制》2015年第9期143-146,151,共5页Fire Control & Command Control

基　　金：国家自然科学基金资助项目(61370031)

摘　　要：发射系统是地空导弹武器系统的重要组成部分,研究发射系统的故障诊断,可提高地空导弹武器系统的作战效能和部队的快速反应能力。BP神经网络在故障诊断方面收敛速度慢、易于陷入局部极小点。为解决上述问题,以液压系统的柱塞泵为例,提出一种蚁群算法改进BP神经网络的故障诊断的方法,优化神经网络的权值和阈值,使网络具有全局兼局部寻优能力。实验结果表明,蚁群神经网络比BP神经网络收敛速度快,运算效率高,识别能力强,并且提高了诊断的准确性和可靠性,是一种有效可行的故障诊断方法,具有良好的应用效果。Launch system is an important part of surface to air missile weapon system,and it is of great significance to research its fault diagnosis on improving the operational effectiveness of air defense missile system and rapid response capability. BP neural network is easy to fall into local minimum point and the convergence speed is slow in fault diagnosis. In order to overcome these shortcomings,this paper introduces ACO algorithm into BP neural network to optimize the thresholds and weights taking the plunger pump in the hydraulic system as an example. Therefore the probability of training algorithm to converge to global optima is improved. The experimental results show that ACO neural network have faster convergence speed,higher efficiency and recognition ability than BP neural network, and it effectively improves the accuracy and efficiency of fault diagnosis. It is a kind of effective and feasible method and has good application prospects.

关 键 词：发射系统 蚁群算法 神经网络 柱塞泵 故障诊断 

分 类 号：TJ768[兵器科学与技术—武器系统与运用工程]