Abstract: At present, most of the domestic service B737NG aircraft using CFM56-7 type engine. As the research object, by decoding the vast data of quick access recorder (QAR), the denoising method with stationary wavelet Rigorous SURE, analyzing the engine speed, N1 speed, N2 speed, exhaust gas temperature (EGT)and other parameters, processing linear or nonlinear regression analysis, combined with the flight phase rational division and modeling, analysis of the main performance parameters and aircraft fuel flow (FF) relationship, full range of the FF prediction model is established. Through MATLAB-Simulink simulation analysis, 5 representative cases of the long flight, short-middle flight, take off-go around flight and complex flight are selected, and errors of the FF prediction model and the actual flow are compared to prove that the model is reasonable and universal.

Key words: QAR data, fuel flow prediction, regression analysis

摘要： 以目前国内服役最多的B737NG飞机所使用的CFM56-7发动机为研究对象，对飞机快速存取记录器（QAR）的大量数据进行译码分析，以平稳小波Rigorous SURE的方法对数据进行预处理、滤波去噪，分别对发动机N1转速、N2转速、排气温度EGT等参数进行线性或非线性回归分析，结合飞行阶段进行合理划分与建模，分析研究出飞机各主要性能参数与燃油流量（FF）的关系，建立FF的全航程预测模型。通过MATLAB-Simulink进行仿真分析，并选取长航班、中短途航班及复飞航班等5种具有代表性的情况验证FF预测模型与实际流量的误差均在允许范围内，证明所建模型的合理性和普适性。

关键词: QAR数据, 燃油流量预测, 回归分析