摘要
以S形进气道为研究对象的主动流动控制研究中,流场状况分析对控制器的设计起到至关重要的作用,而在实时控制中,显然不可能通过流场的数值模拟获得流场分布情况。本研究从神经网络模型辨识理论出发,结合进气道流场的数值模拟结果和实验采集数据,对神经网络进行训练和验证,建立了不同来流马赫数下进气道沿程壁面静压的预测模型,模型拟合误差为0.007,预测结果与实际实验结果相符,证明了从辨识理论出发建立流场模型的可行性,为流场状况的实时获取提供了可靠易行的方法。
In the study of active flow control with S-shaped inlet duct as the research object,the analysis of flow field plays an important role in the design of the controller.In the real-time control,it is evidently and practically impossible to have the flow field distribution through numerical simulation.Based on the neural network model identification theory,and the numerical simulation and experimental data of the inlet duct flow field,the neural network is trained and validated,and then the prediction model of the static pressure of the inner wall along the inlet duct with different mach numbers is established.It proves the feasibility of establishing the flow model from the identification theory,providing a reliable and easy method for the real-time acquisition of the flow field.
出处
《热能动力工程》
CAS
CSCD
北大核心
2017年第11期8-12,共5页
Journal of Engineering for Thermal Energy and Power
关键词
流场
辨识
预测模型
神经网络
flow field
identification
prediction model
neural network
