基于AGA-RS的故障特征参数提取方法研究  

Method research of fault feature extraction based on AGA-RS

作　　者：李华莹[1] 刘建敏[1] 乔新勇[1] 李晓磊[1] 

机构地区：[1]装甲兵工程学院机械工程系,北京100072

出　　处：《计算机工程与应用》2014年第5期257-260,共4页Computer Engineering and Applications

摘　　要：在故障诊断中,将高维特征空间压缩到低维特征空间可以简化故障分类器设计,提高运算效率。研究了自适应遗传算法(AGA)和粗糙集(RS)理论在特征选择和特征约简中的应用,并针对柴油机燃油喷射系统故障提取了简化特征,建立了神经网络模型。试验结果表明,基于AGA-RS的故障特征参数提取方法可使故障分类器输入参数同时具有有效性和简约性,提高了神经网络的运算效率。In fault diagnosis, compressing high dimension feature space to low dimension feature space can simplifiy the design of fault classifier and improve computational efficiency. Adaptive genetic algorithm and rough sets theory are researched for feature selection and reduction, the brief features are abstracted according to the faults of diesel fuel injection system, and the neural network model is built. Experimental results indicate the method can extract effective and brief features, but also improve the capacity of neural network.

关 键 词：自适应遗传算法 粗糙集 故障诊断 特征提取 

分 类 号：TK423[动力工程及工程热物理—动力机械及工程]