BP神经网络在风洞模型静态测压试验结果预测中的应用研究

Application of BP Neural Network in Forecasting Static pressure test results of Wind Tunnel Model

通过飞行器模型风洞静态测压试验,可以得到飞行器的静态性能参数,但在具体的试验过程中,由于现场条件所限仅能够得到特定试验状态下的有限的数据。为了进行更为全面的研究,文章采用BP神经网络算法,利用已测状态的试验数据,设计一个用于数值预测的BP神经网络,进行两方面的预测,一方面是在确定模型迎角的前提下,预测模型不同测压孔位置处的压力值;另一方面是在确定模型测压孔位置的前提下,预测模型不同迎角下的压力值。最终将预测结果与实际结果进行对比,两者的相对误差小于±2%。结果表明:在有限的试验状态下,BP神经网络对风洞模型静态试验结果的预测较为准确,符合实际的流动特性,可以为飞行器设计提供较为全面、可靠的试验数据。

Through the static pressure measurement test of the aircraft model wind tunnel,the static performance parameters of the aircraft can be obtained,but in the specific test process,due to the field conditions,only limited data can be obtained under the specific test state.In order to carry out more comprehensive research,this paper uses BP neural network algorithm,uses the measured state of the experimental data,designs a BP neural network for numerical prediction,and carries out two aspects of prediction.On the one hand,under the premise of determining the angle of attack of the model,the pressure value at different positions of the pressure hole in the model is predicted;on the other hand,under the premise of determining the position of the pressure hole of the model,the pressure value of the model at different angles of attack is predicted.Finally,the relative error between the predicted results and the actual results is less than±2%.The results show that the prediction of static test results of wind tunnel model by BP neural network is more accurate and in line with the actual flow characteristics under limited test conditions,and can provide more comprehensive and reliable test data for aircraft design.