固定翼无人机健康退化建模方法研究

Research on health degradation modeling of fixed-wing UAVs

固定翼无人机随使用时间累积,健康状态会逐步退化。研究其健康退化建模方法,可用于无人机剩余寿命预测,指导开展预先维修、智能维修,从而有效延长装备使用寿命和减少故障发生概率。针对固定翼无人机全寿命数据难以获取及故障数据稀缺导致建模困难的问题,提出了基于多任务批次健康飞行数据的建模方法。该方法以SVDD算法为基础,构建无人机健康超球体模型,以球心距偏移趋势刻画健康退化程度,在此基础上引入KDE算法构建飞机整体健康退化指标,进一步使用LSTM算法实现多批次退化趋势的拟合。该方法可应用于固定翼无人机维护与维修,减少维修成本。

The health status of Fixed-wing UAVs will gradually deteriorate with its operation time.The research on its health degradation modeling method can be used to predict the remaining life of UAV,guide the advance maintenance and intelligent maintenance,thus effectively extend the service life of equipmentand reduce the probability of failure.In this paper,a modeling method based on multi task batch healthy flight data is proposed,which is used to solve the problem that the life cycle data of Fixed-wing UAVs is difficult to obtain and the scarcity of fault data.This method is based on the Support Vector Data Description(SVDD)algorithm,and it constructs a health hypersphere model of UAV,characterizes the degree of health degradation with the trend of spherical distance offset,introduces the Kernel Density Estimation(KDE)algorithm to construct the overall health degradation index of UAV,and further uses the Long Short-Term Memory(LSTM)algorithm to achieve the fitting of multi batch degradation trends.The method is expected to be applied to the maintenance and repair of Fixed-wing UAVs and reduce the maintenance cost.