基于UKF的共轴式无人直升机模型辨识

建立了共轴式无人直升机系统非线性模型,并针对其非线性强,不同飞行模态下气动参数差异等问题,将无迹卡尔曼滤波(UKF)引入共轴式直升机系统非线性模型辨识,不但避免了直升机线性模型仅仅适用于悬停模态的局限性,同时为直升机系统在线自适应控制提供了基础条件,使得共轴式无人直升机自主全包线飞行成为可能.以北京航空航天大学FH-1共轴式无人直升机为例进行了仿真辨识实验.实验结果表明基于该方法的共轴式直升机在线非线性模型辨识不依赖于参数初值的选取,模型参数能在10s内收敛,各状态量辨识精度达到80%以上,明显高于传统的预报误差法(PEM),具有一定的实用性.

Identification of unmanned coaxial helicopter model based on UKF

Nonlinear model of the unmanned coaxial helicopter system was built, and on account of considering its strong nonlinear character, as well as the aerodynamic parameters were variable under different flight modes, the unscented Kalman filter (UKF) was introduced to solve the nonlinear model identification problem of coaxial helicopter. It did not only avoid the limitations that linear model was only appropriate to hover modes of helicopter model, but also provided the basis for the online adaptive control of helicopter system, in which autonomous unmanned coaxial helicopter's full envelope flight could be possible. Identification of the FH-1 unmanned coaxial helicopter developed by Beijing University of Aeronautics and Astronautics was simulated by the approach and the predictive error method(PEM). Simulation experiment results show that the online identification of coaxial helicopter nonlinear system based on UKF does not depend on the selection of initial parameters, parameters can converge within the validity period of 10s, and the accuracy of identification reached 80%,which is higher than the classical PEM, so it has a certain practicality.