摘要
为降低装备全寿命周期费用、提高经济可承受性、预防灾难性事故的发生,开展了航空电子设备故障预测技术研究;采用粗糙集理论改进神经元结构,以粗糙变量为神经元的输入,每个神经元的上近似元和下近似元分别代表粗糙集的上下近似,以交叉连接方式构造粗糙神经网络,用以实际设备特征参数的跟踪预测;研究结果表明,粗糙神经网络可以较准确地预测故障发生的时间,且较BP神经网络预测性能有较大改善;该方法对于航空电子设备的维护保障具有一定的理论价值和现实意义。
In order to cut down the equipment life cycle cost, improve affurdability and prevent serious accident, the research of avionics prognostics is carried out. The structure of neurons was ameliorated with the Rough Set Theory, and the Rough Neural Network （RNN） was got to track the degenerate trend of the characteristics parameters of practical equipment. The result of research shows that Rough Neural Network can forecast the fault in high veracity, and have a better predicting performance than BP neural network. This prognostics method contributes to the level of the avionics maintenance both in theory and activity.
出处
《计算机测量与控制》
CSCD
北大核心
2010年第4期807-809,共3页
Computer Measurement &Control
基金
"十一五"国防预研项目(51317030105)
