基于RBF神经网络的某复杂装备故障预测方法  

Predication Method of Complex Equipment Based on RBF Neural Network

作　　者：黄波[1] 丁浩[1] 张孝芳[1] 衡辉[1] 

机构地区：[1]海军潜艇学院导弹兵器系,山东青岛266042

出　　处：《计算机仿真》2014年第1期14-17,共4页Computer Simulation

基　　金：军内科研(装计2012(475))

摘　　要：某复杂装备的工作状态直接影响着部队的战斗力,对其进行科学的故障预测尤为重要。针对装备的非线性和复杂性,提出RBF神经网络模型的故障诊断和预测方法,确保了故障预测的准确度。利用神经网络的非线性建模能力,在装备的关键监测点建立故障诊断器,通过神经网络的训练学习,确定需要的参数估计,根据模型的输出值来判断故障。仿真结果表明,改进方法的预测结论与实际情况基本一致,可为解决同类问题提供有价值的借鉴。The running state of a complex equipment has a direct impact on battle effectiveness, so scientific fault predication is especially important for it. Considering the equipment has non-linear and complexity, a fault detection and fault predication method based on RBF neural network was proposed, which possesses higher accuracy. A neural network was constructed by non-linear capability of model making , and a network diagnosing model was constructed by a set of measuring points in the equipment. The parameter estimations were given by learning of neural network, and the fault was judged by the actual output of model. The simulation result shows that the predication result is basi- cally concordant with the actual situation, and the method provides a valuable reference for homologous problem.

关 键 词：径向基函数 复杂装备 故障预测 

分 类 号：TP391.5[自动化与计算机技术—计算机应用技术]