嵌入式大气数据系统Kriging算法模型

A Novel Flush Airdata System Model Based on Kriging Algorithm

基于Kriging算法提出了一种嵌入式大气数据系统的计算模型。该模型以测压点压强为输入,攻角、偏航角和Ma等大气参数为输出。采用Kriging插值方法,建立了输入与输出间的映射关系,从而避免了迭代类方法反复迭代的过程,实现了高实时性。分别比较了取三、四和五个测压点时,以及取不同位置处测压点时,计算大气数据的精度。通过与神经网络模型比较,Kriging模型的精度明显优于神经网络模型。

A novel numerical model which calculates the air data for flush airdata system is developed based on Kriging algorithm. The pressure measured by the sensor is taken as inputs, and the atmospheric parameters such as angle of attack, angle of slide and Ma are taken as outputs. By using Kriging interpolation method, the mapping between the input and the output is established, the interative process of interation method is avoided and the air data can be calculated in real-time. The accuracy of the model is compared with the neural network model with different numbers of inputs, and the positions of the sensor are also taken into account in the com- parison. The accuracy of Kriging model is superior than the neural network model.