【作者】 赵磊；

【导师】 林晓辉；

【作者基本信息】 哈尔滨工业大学 ， 飞行器设计， 2008， 硕士

【摘要】 飞行器气动参数辨识研究的主题,是应用系统辨识技术从飞行试验数据求取气动力,从而建立飞行器动力学系统的数学模型。它研究的对象是飞行器,解决的是空气动力学问题,采用的基本方程是飞行动力学的运动方程组,应用的研究手段是现代控制论中的滤波、预测和估计理论。本文在纵向非线性动力学模型已知的情况下,用参数估计的方法,辨识飞行器运动过程中未知的气动力系数。首先,针对基于极大似然法和广义卡尔曼滤波的参数辨识方法,研究了飞行器气动参数辨识的一般特性。然后,从辨识的精确性和效率性出发,针对计算精度高、计算效率低的极大似然一般算法采用了极大似然近似算法降低了极大似然法在辨识过程中计算效率低的问题;给出更高的辨识效率算法增广的广义卡尔曼滤波算法,把状态方程中的待辨识参数做为状态量加入状态方程中,利用广义卡尔曼滤波进行状态估计求解待辨识参数;针对增广的广义卡尔曼滤波计算效率高计算精度差的情况,提出增广的拟线性最优平滑滤波法对状态方程进行高度线性化处理,从而减少卡尔曼滤波中状态方程线性化不足所带来的误差,在精算量增加不大的前提下,辨识精度有了明显的提高。通过增广的拟线性最优平滑滤波法,解决了极大似然法和增广的广义卡尔曼滤波在辨识过程中计算精度和计算效率的矛盾。最后,基于遗传算法的随机搜索算法,利用其全局优化特点,针对三种辨识方法建立适应度函数,对三种辨识算法所对应的个体进行评价,求解出最优权系数,组合优化从而进一步得到更精确的辨识解。更多还原

【Abstract】 The theme of vehicle aerodynamic parameters identification applicated system identification technologies got the aerodynamics from flight test data to establish a mathematical model of dynamic systems. It is the object of research aircraft to solve the aerodynamics problems, the equation is the flight dynamics of the campaign equations, application of modern means of control theory of filtering, projections and estimates theory.This paper has been aware from the vertical nonlinear dynamics model, the estimated parameters of the method used to identify unknown aircraft in the course of the campaign aerodynamic coefficients.First of all, this paper based on the Maximum Likelihood (M.L) algorithm and Extended Kalman Filter(E.K.F) research the vehicle identification of the general parameters of aerodynamic characteristics.The problem of Maximum Likelihood algorithm for calculating the efficiency of low and high accuracy to Extended Kalman Filter high-efficiency problem of low accuracy afford the Approximation Maximum Likelihood algorithm and Extended Quasi-linear Optimum Smoothing Filter to reduce the Maximum Likelihood algorithm in the identification process in terms of low efficiency and the low calculating the efficiency of EKF. Solve the conflict of Maximum Likelihood algorithm and Extended Kalman Filter in the process of identification accuracy and efficiency .Finally, based on genetic algorithm of random search algorithm, using its global optimization characteristics of different identification method to establish fitness function, corresponding to different identification algorithm to evaluate the individual, for the optimal weights, so as to optimize the combination of more accurate Identification solution.更多还原