飞机NACA进气口性能的数据驱动预测模型研究

Research on Data Driven Prediction Model of Aircraft NACA Intake Performance

冲压空气进气口是民机环控系统性能实现的关键。针对飞机不同结构参数的NACA进气口,提供三种数据驱动的预测模型,对其性能进行预测。分析多元线性回归模型、二阶多项式模型、人工神经网络三种预测模型的原理,建立相应的数学模型,并在MATLAB实现三种预测模型的代码编译。以某一型号客机为例,进行ANSYS CFX流场仿真,获取不同结构参数对应性能指标的数据库,基于数据库对比不同预测模型下的误差,并分析出适用于飞机NACA进气口不同结构参数性能预测的模型。对较优的BP神经网络模型进行改进,得到更加适合的BP神经网络改进模型。

Ram air inlet is critical for realizing the performance of civil aircraft environmental control system.This paper provides three data driven prediction models for NACA air inlet with different structural parameters of aircraft to predict its performance.This paper analyzes the principles of the three prediction models,which are multivariate linear regression model,second order polynomial model and Artificial Neural Networks,establishes the corresponding mathematical model,and implements the code compilation of the three prediction models in MATLAB.Taking a certain type of civil aircraft as an example,the simulation based on ANSYS CFX was carried out to obtain the database of corresponding performance indexes of different structural parameters.Based on the database,the errors of different prediction models were compared,and which model is suitable for predicting the performance of NACA inlet with different structural parameters was analyzed.Finally,this paper improves the better BP neural network model and obtains the more suitable BP neural network improved model.