考虑数据选择的航空发动机基线建模方法

Aircraft Engine Baseline Modeling Method Based on Data Selection

航空发动机基线是发动机状态监控的基础,当前国内使用的基线大部分都由发动机制造商所提供。近年来,为打破发动机制造商对基线的技术封锁,从飞机通信寻址与报告系统提供的数据中挖掘基线的方法受到了广泛的研究。但是,当前大部分关于基线建模的研究忽略了建模数据的影响。因此,本文提出考虑数据选择的航空发动机基线建模方法。该方法基于多元线性回归建立基线模型,并通过网格搜索确定最优的建模数据选取方案。以建立排气温度基线为例,该方法在某航空公司提供的CFM56-7B发动机巡航报上进行试验。结果表明,使用发动机开始服役2201个飞行循环后的4405个飞行循环的数据进行基线建模可以得到最小的建模误差。

Aircraft engine baseline is the key foundation of engine condition monitor.The baseline is usually provided by engine original equipment manufacture.In recent years,there are lots of researches on baseline modeling method based on airlines data.However,most of current researches focus on model selection and optimization,and ignore the modeling data factor.In this paper,a baseline modeling method that considers the modeling data is proposed.The proposed method develops baseline by using multi-liner regression and obtains the best modeling data selection by using grid searching.The proposed method is verified in CFM56-7B cruising data collected from an airline.The results show that the optimal baseline is obtained by using 4405 cycles after engine is commissioned 2201 cycles.