Abstract: As a non-linear, multi-variable and highly coupled underactuated system, the quadrotor UAV's underactuated characteristic can cause instability of the quadrotor nonlinear link, and then cause interference to the strapdown inertial measurement system. To solve this problem, first analyze the structure and characteristics of the quadrotor underdrive system, and on the basis of which establish the Hammerstein nonlinear model of the quadrotor UAV system by reference to the composition of the traditional Hammerstein nonlinear system. Then we design a new Kalman-type scroll window complementary fusion filtering algorithm and verify the performance of the algorithm on the underactuated Hammerstein system of quadrotor drone. Simulation and physical test results show that the fusion filtering algorithm has good smoothness and fast followability. It can minimize the interference of external noise on the quadrotor flight control under the premise of ensuring the system response speed, and effectively solve the validity and stability issue of the data collected by the strapdown inertial measurement system under complex conditions.

Key words: Kalman type rolling window, complementary fusion filter, underactuated, Hammerstein system

摘要： 四旋翼无人机作为一个非线性、多变量、高度耦合的欠驱动系统，其欠驱动特性会引起四旋翼非线性环节不稳定，进而对捷联惯性测量系统产生干扰。针对这一问题，首先分析了四旋翼欠驱动系统的结构和特征，参考传统Hammerstein非线性系统的组成，建立四旋翼无人机系统的Hammerstein非线性模型；然后设计一种新的卡尔曼式滚动窗口互补融合滤波算法，并在四旋翼无人机欠驱动Hammerstein系统之上验证算法的性能。仿真和实物测试结果表明，该融合滤波算法具有良好的平滑性和快速的跟随性，能够在保证系统响应速度的前提下，最大程度削弱外界噪音对四旋翼飞行控制的干扰，有效解  决了复杂状况下捷联惯性测量系统采集数据的有效性和稳定性问题。

关键词: 卡尔曼式滚动窗口, 互补融合滤波, 欠驱动, Hammerstein系统