采用改进NMPC方法的气浮卫星模拟器控制研究

Research of Air Float Satellite Simulator Using Improved NMPC Approach

随着航天技术的发展,微纳卫星在交会对接、绕飞等任务中的应用越来越广泛。然而,微纳卫星在执行上述任务时,对控制精度有较高需求,同时星载平台功耗低、算力有限,难以部署优化控制方法。针对这一问题,开展了微纳卫星近距离抵近绕飞对接的地面仿真控制研究。首先,设计了一套地面气浮台仿真系统,包含配备八个微推力器的气浮卫星模拟器,能够实现实时姿轨协同控制。其次,针对直线抵近、凝视绕飞和交会对接等机动任务,采用基于约束感知粒子滤波的改进模型预测控制器建立优化问题,设计了推力饱和等多种性能约束以及姿态指向等任务约束。随后,通过蒙特卡罗采样方法对非线性优化问题进行求解,避免了传统非线性模式预测控制易于局部收敛的问题。最后,通过仿真与实物验证,对所设计的控制算法进行了性能测试。实验结果表明,所设计的优化控制算法对计算资源的消耗较低,可在低功耗星载计算机上以2 Hz实时输出控制指令,该方法可以实现对目标的精确抵近、绕飞及高精度对接,其中位置控制误差为0.0078 m,姿态控制误差为0.62°,与传统控制算法相比精度得到了明显提升。该研究为星载平台部署优化控制算法提供了一种有效解决方案,具有较高的工程应用价值。

With the development of space technology,micro/nano satellites are more and more widely used in rendezvous and docking,fly-around and other tasks.However,micro/nano satellites have a high demand for control accuracy when performing the above tasks,while the low power consumption and limited arithmetic power of the on-board platform make it difficult to deploy optimized control methods.Aiming at this problem,this paper carries out a ground simulation control study of micro/nano satellite near-arrival around flight docking.First,a set of ground air float platform simulation system is designed,including air float satellite simulator equipped with eight micro-thrusters,which can realize real-time attitude-orbit cooperative control.Second,for maneuvering tasks such as straight-line approach,fly-around and rendezvous and docking,a variety of performance constraints such as thrust saturation and mission constraints such as attitude pointing are designed by using the Constraint-Aware Particle Filtering-based Modified Model Predictive Controller(CAPF-NMPC).Subsequently,the nonlinear optimization problem is solved by Monte Carlo sampling method,which avoids the problem of easy local convergence of traditional NMPC.Finally,the performance of the designed control algorithm is tested through simulation and physical verification.The experimental results show that the designed optimization control algorithm consumes less computational resources and can output control commands in real time at 2 Hz on a low-power on-board computer,and the method can achieve precise approach,orbiting and high-precision docking of the target,with a position control error of only 0.0078 m and an attitude control error of 0.62°,significantly improves accuracy compared to conventional control algorithms.This study provides an effective solution for the deployment of optimized control algorithms for on-board platforms,which has high engineering application value.