基于BP神经网络的燃气流量可调固体冲压发动机推力建模方法研究  

Research on Thrust Modelling Method of Variable Flow Ducted Engine Based on BP Neural Network

作　　者：周宇宸 曾庆华[1] 王安 Zhou Yuchen;Zeng Qinghua;Wang An(College of Aerospace Science and Engineering,National University of Defense Technology,Changsha 410073,China)

机构地区：[1]国防科技大学空天科学学院,长沙410073

出　　处：《航空兵器》2018年第4期28-32,共5页Aero Weaponry

摘　　要：针对流量可调固体冲压发动机稳态数值模型建立的问题,提出了一种基于BP神经网络的建模方法。该方法利用数值计算与试车台试验数据相结合的方法,得到基于高度、马赫数、迎角、燃气发生器压强、补燃室压强五个关键特征参数的推力数据库,基于该数据库建立了多层BP神经网络模型,并对该模型进行了校核验证。结果表明:所建立的多层BP神经网络模型可精确预测理论数值计算模型,可代替复杂的数值计算模型或大规模的推力数据库,为固体冲压发动机弹上控制系统产品的研制奠定了基础。To solve the problem of steady mathematical modelling of variable flow ducted engine, a modelling method based on BP neural network is proposed. Combining numerical calculation with test data of variable flow ducted engine, the thrust database based on five key characteristic parameters, including height, Math number, angle of attack, pressure of gas generator and pressure of afterburining chamber, is obtained. Based on this database, a multilayer BP neural network model is established, and the model is checked and verified. The results show that the establised model can accurately predict the theoretical mathematical model of variable flow ducted engine, and replace complex mathematical model or largescale thrust database in some situations, which lays a foundation for the development of missile control system products of ducted engine.

关 键 词：流量可调 固体冲压发动机 BP神经网络模型 建模方法 

分 类 号：TJ763[兵器科学与技术—武器系统与运用工程] V435[航空宇航科学与技术—航空宇航推进理论与工程]