某型军用飞机下降阶段燃油消耗模型研究

Research on Fuel Consumption Model of a Military Aircraft in Descent Stage

摘要: 针对飞机下降阶段燃油消耗模型建立不够精准的问题，提出了一种改进的长短期记忆网络（Long Short-Term Memory, LSTM）方法，基于飞参数据建立了某型军用飞机下降阶段的燃油消耗模型。首先从飞参数据中提取与燃油消耗相关的参数并进行预处理，然后利用互信息的方法筛选了与燃油流率高度相关的参数，最后基于一种改进的LSTM网络建立了飞机下降阶段的燃油消耗模型。该法相比于传统的BP神经网络、回声状态网络以及标准LSTM在精度上都有了较大的提升和改进。

Abstract: Aiming at the problem of inaccurate establishment of fuel consumption model in aircraft descent stage, an improved Long Short-Term Memory (LSTM) network method is proposed. Based on flight data, the fuel consumption model of a military aircraft in descent stage is established. Firstly, the parameters related to fuel consumption are extracted from flight data and preprocessed. Then, the parameters which are highly correlated with fuel flow rate are selected by Mutual Information method. Finally, a fuel consumption model of aircraft descent stage is established based on an improved LSTM network. Compared with traditional BP neural network, Echo State Network and standard LSTM, this method has a greater improvement in accuracy.