基于改进凌日搜索算法的风洞天平载荷预测方法

A Wind Tunnel Balance Load Prediction Method Based on Improved Transit Search Algorithm

摘要:

风洞天平是在风洞测试中使用的测力传感器，在使用之前需要进行校准以测量缩比模型受到的气动载荷；传统的方法使用预设的多项式函数进行拟合，忽略了某些变量的存在对测量载荷的负面影响，导致数据处理结果的失真；在此，提出了一种改进的凌日搜索算法(ITS)，选择对测量载荷更具重要性的特征；然后，使用贝叶斯线性回归算法(BLR)建立预测模型测量天平载荷，最后，在两个天平数据集上测试了该方法，结果表明ITS-BLR方法评估确定了对预测目标具有较高贡献的特征，进而降低了预测误差，与最小二乘法得到的综合加载误差相比，降幅最高达到60%，说明提出的方法可以实现对天平载荷的准确预测。

Abstract:

The wind tunnel balance is a force sensor used in wind tunnel testing, which needs to be calibrated to measure the aerodynamic load on the scale model before it is used. The traditional method uses the preset polynomial function for fitting, ignoring the negative influence of some variables on the measured load, which leads to the distortion of data processing results. In this paper, an improved transit search algorithm (ITS) is proposed to select the features that are more important for load measurement. Then, Bayesian linear regression algorithm (BLR) was used to build a prediction model to measure the balance load. Finally, the method was tested on two balance datasets. The results showed that the ITS-BLR method evaluated and identified the features with higher contribution to the prediction target, thereby reducing the prediction error. The reduction is up to 60%, which shows that the proposed method can accurately predict the balance load.