粗糙集约简在飞行器故障识别中的应用  

Application of Rough Set Reduction in Spacecraft Fault Diagnosis

作　　者：赵方[1] 王静[1] 杨海龙[2] 

机构地区：[1]北京林业大学信息学院,北京100083 [2]航天东方红卫星有限公司,北京100094

出　　处：《计算机科学》2012年第B06期429-431,456,共4页Computer Science

摘　　要：故障识别是飞行器结构健康监测的重要内容,不同类型的结构测量数据之间存在互补和矛盾的成分。粗糙集约简具有良好的数据挖掘和归类能力,为处理多传感器冗余信息提供了有效的工具。首先从飞行器结构的位移、加速度、应变响应以及模态参数中提取综合的故障评价指标,然后利用粗糙集约简技术对特征属性进行约简,求得系统的最简特征集合,最后通过概率神经网络进行了结构故障的识别。直升机结构的损伤仿真算例表明粗糙集约简方法不仅可以显著减低特征属性的维度,而且能提高故障识别的精度。Damage detection of the aerospace structures is a critical issue,during which the data from multiple types of sensors are redundant and inconsistency to each other.Rough set reduction is an effective tool for data mining and classification,which can be used for the above problems.Firstly,the damage features are extracted from the multi-sensor information（e.g.displacement,acceleration or strain sensor） and modal parameters.Then,the rough set reduction technique is employed to obtain the core set of all features.Finally,the structural damage conditions are identified through probabilistic neural network.The numerical simulation of a helicopter demonstrates that the rough set reduction technique can not only decrease the spatial dimension of feature attributions,but also improve the identification accuracy.

关 键 词：粗糙集 飞行器 故障识别 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]