基于飞行试验和风洞试验数据的融合算法研究

Research on the data fusion algorithm based on flight test data and wind tunnel test data

飞行试验受飞行轨迹和试验激励的限制,通过飞行遥测数据辨识得到的实飞气动数据只能分布于飞行轨迹上的少量特征段。风洞试验由于受地面模拟能力的限制,试验数据与实飞飞行数据存在一定的天地差异。为了弥补飞行试验辨识数据和风洞试验数据的“缺陷”,以一种典型轴对称飞行器为研究对象,对上述两种不同来源的气动数据进行数据融合方法研究。首先,通过气动力参数辨识方法得到飞行器沿实际飞行轨迹的六分量真实气动数据,然后比较与地面试验数据的差异性和一致性;最后分别采用基于梯度信息和基于高斯过程回归的数据融合方法对两种来源的气动数据进行融合。融合建模结果表明,两种融合模型预测数据均比单源模型精度更高;如果高、低精度数据的梯度信息较为一致,则基于梯度信息的数据融合方法效果更好;而基于高斯过程回归的数据融合方法能够给出融合数据的置信区间,有利于分析不确定度。

Due to the correlation between flight and ground test data,there is a certain difference between the wind tunnel test data and the flight test data.Limited by the flight trajectory and test excitation,the flight test data distributes in a small characteristic sections of the flight trajectory.And due to the limitation of ground simulation capabilities,wind tunnel tests have certain differences between the test data and the actual flight data.In order to remedy these limitations,researches of the data fusion algorithm based on above two different sources of aerodynamic data are conducted.Firstly,through the aerodynamic parameter identification method,the sixcomponent real aerodynamic data of the aircraft along the actual flight trajectory are obtained,and then compare the difference and consistency with the wind tunnel test data.Finally,two kinds of data fusion methods based on gradient information and Gaussian process regression were used to fuse the two sources of aerodynamic datas.The results show that both the prediction data of the two fusion models are more accurate than the single source model;if the gradient information of the high and low precision data is more consistent,the data fusion method based on the gradient information is more effective;while the data fusion method based on the Gaussian process regression could obtain the confidence interval of the fusion data,which is useful to the analysis of uncertainty.