基于 Stacking集成学习的飞行模拟器设备故障预判

为精准研判飞行模拟器故障问题,提升飞行模拟器故障应急处理能力,提出了 1种基于 Stacking集成学习的设备故障预判方法。首先,针对设备故障数据具有高维度和时序性的特点,采用正态分布法,对采集的原始故障数据进行缺失值和异常值处理,获取归一化数据,并基于核主成分分析法,提取数据的特征并降低数据维度；然后,构建以 CNN、RF、BP神经网络为基分类器,XGBoost为元分类器的 Stacking集成学习模型,采用交叉验证训练策略,对设备的故障类型作出预判；最后,构架典型案例。仿真结果表明,所提方法可正确有效地识别设备的多种故障状态。

Fault Prejudgment of Flight Simulator Equipment Basedon Stacking EnsembleLearning

Inordertopredictfaultstatesofflightsimulatoraccuratelyandeffectively,andimprovetheabilitytodealwith emergenciesoftheflightsimulatorequipmentfault.Theequipmentfaultprejudgmentmethodbased onStackingensem-blelearningisproposed.Thenormaldistributionmethodisused toprocessthemissingvaluesandoutliersofthecollect-edoriginalfaultdata toobtainthenormalizeddataforthecharacteristicsofequipmentfaultdatawithhighdimensionlity andtime-series.Basedonkernelprincipalcomponentanalysis,thefeaturesofdataareextractedandthedimensionsofda-taarereduced.AStackingensemblelearningmodelwithCNN,RFandBPneuralnetworksasthebaseclassifierandXG-Boost as the meta-classifier is constructed, and a cross-valiclation training strategy is used to predict the fault types of equipment.Typical casesareconstructedandsimulationresultsshowthattheproposedmethod canaccuratelyandeffec-tivelyidentifyvariousfaultstatesofequipment.