直升机非线性模型结构和参数鲁棒辨识方法

Robust identification method for nonlinear model structures and parameters of helicopter

提出了一砷能有效地对直升机模型进行辨识的鲁棒选择模型的新方法.通过分析数据矩阵模型判定方法,采用U-D分解方法以避免行列式的复杂计算,提高了计算效率.通过估计D阵元素的取值区间,得到了观删量不确定部分带来的模型辨识判据的误差上下界.依此判据,将侯选模型按照重要程度逐个选取.参数估计采用了加权最小二乘算法,按照上界不等式对估计误差的协方差阵进行了估计,得到了鲁棒辨识的新算法.为了便于工程应用,给出了鲁棒辨识算法的收敛条件.根据飞行试验结果对直升机悬停、前飞模型和参数进行辨识的结果表明,新方法可以得到工程上满意的结果.

To identify the model structures of helicopter efficiently, a new robust criterion for model structure selection is developed. By the analysis of model determination method of sample data matrix, a numerical excellent U-D factorization is presented to avoid the computation of determinant values of two complex matrices and improve the computational efficiency. By means of estimating, the values of matrix D, both the upper bound and the lower bound of the influences to model criterion caused by system uncertainty, is described. To get robust algorithm of parameter estimation, the lower bound of uncertain parts generated by uncertain observation matrix in the expression of the error covariance matrix of parameter estimation is given, Thus, both model structure and parameter identification algorithms are suggested. Results of forward and hovering data processing for both longitudinal and lateral flight tests of a helicopter show that this method is efficient and satisfactory result in engineering may be obtained.