基于解析模型的小样本操纵性KT指数辨识方法

Analytical Model for Maneuverability KT Index Identification with Small Sample

在研究KT操纵响应模型的基础上,求解在连续线性操舵情况下一阶和二阶操纵响应解析模型。直接利用航向角数据构建目标函数,采用基于格雷码和精英选择的遗传算法进行非线性寻优。通过与最小二乘法比较,结果表明该算法能将航向角平均偏差降低约40%。在辨识过程中,调整航向角的时间步长,进行多次辨识验证后得出该方法只需要小样本数据就可进行高效准确的辨识,极大地提升辨识效率。利用不确定度计算原理,在样本存在一定误差的前提下,有效证明该辨识方法的可靠性。

Based on the KT maneuvering response model, an analytical model for the first-order and second-order maneuvers in the case of continuous linear steering is developed. The objective function is constructed directly by the heading angle data, and the nonlinear optimization is performed by the genetic algorithm based on Gray code and elite selection. Compared with the least squares method, it is proved that the algorithm can reduce the average deviation of the heading angle by about 40%. The identification process is verified with different sampling time step of the heading angle. Experiments show that the method achieve efficient and accurate identification with small sample data. Using the uncertainty calculation principle, the reliability of the identification method is proved reliable, even if the sample has certain error.