一种基于机载数据的民用航空器飞行阶段划分方法

A flight phase identification method based on airborne data of civil aircraft

为有效解决民用航空器的机载快速存取记录器(QAR)航迹数据中飞行阶段出现错误划分的情况,通过航空器气动构型和垂直运动态势变化提出了飞行阶段重新划分方法,具体分为4个步骤:数据预处理、垂直运动态势划分、飞行状态特征模型构建和飞行阶段重新划分;使用基于DBSCAN的局部遍历聚类方法对气压高度变化趋势进行聚类分析,划分垂直运动态势,并通过设置有效态势最短持续时间解决气压高度的局部抖动问题;考虑不同机型在不同飞行阶段时的襟翼位置和操纵各不相同,场面滑行时的场压抖动以及空速表在低速滑行时数值不准确等因素,根据航空器襟翼位置开关、起落架位置、地速和垂直态势等状态参数建立适合所有机型的飞行状态特征模型,并对QAR航迹划分飞行状态特征段;建立各飞行阶段与飞行状态特征的关系模型,结合起落架空地逻辑将所有飞行状态特征段识别为对应的飞行阶段。以3个典型的样本航班为例,计算结果表明:航班包含复飞在内的所有飞行阶段均被有效识别和划分,并与航空器飞行时的襟翼和起落架状态保持一致,原QAR数据中的飞行阶段错误划分问题得到有效解决;对272268个航班的QAR航迹进行了飞行阶段重新划分,成功率为99.7%,起飞、初始爬升、进近和着陆等非光洁构型飞行阶段的平均历时分别为0.6、1.9、6.1和4.0 min,平均距地高度分别为54、3680、6030和2500 ft,符合航班实际运行规律。可见,所建立的飞行阶段重新划分方法可适用于大批量航班,为民用航空器的飞行阶段特征分析提供技术支撑。

To resolve the flight phase identification errors in the airborne quick access recorder(QAR)trajectory data of civil aircraft,a method of flight phase re-identification was proposed based on the aerodynamic configuration and vertical movements of aircraft.This method comprises four steps:preprocessing of QAR data,identification of vertical movements,construction of a flight state characteristics model,and re-identification of flight phases.A DBSCAN-based local traversal clustering method was used to cluster the trends of pressure altitude to classify vertical movements,and a valid minimum state persistence time was set to eliminate the local flutter in the pressure altitude data.Considering aircraft-to-aircraft differences in flap position and control,fluctuations in airfield QFE,and inaccuracies in the airspeed indicator during low-speed taxiing,a flight-state characteristics model suitable for all types of aircraft and based on state parameters,such as flap switch position,landing gear position,ground speed,and vertical movements,was constructed.The model was used to divide the QAR data into flight state characteristic segments.The relationship model between each flight phase and flight state characteristics was established,and all flight state feature segments were identified as corresponding flight stages combined with landing gear air-ground logic.Three typical sample flights are used as examples,calculation results show that all the flight phases(including go-arounds)are correctly identified and divided,and are also fully consistent with the flap and landing gear states of the aircraft.The flight phase identification error in raw QAR data fields is solved effectively.The flight phases of QAR tracks of 272268 flights are re-identified,and the success rate is 99.7%.The average durations of non-clean configuration flight phases,such as take-off,initial climb,approach,and landing,are 0.6,1.9,6.1 and 4.0 min,respectively,and the average altitudes from the ground are 54,3680,6030 and 2500 ft,respectively,which are consistent with the actual flight operation behaviors.Therefore,the flight-phase re-identification method can be applied to numerous flights and provide technical support in analyzing the characteristics of civil aircraft flight phases.5 tabs,8 figs,25 refs.