The state estimation for relative motion with respect to non-cooperative spacecraft in rendezvous and docking(RVD) is a challenging problem. In this paper, a completely non-cooperative case is considered, which means that both orbit elements and inertial tensor of target spacecraft are unknown. By formulating the equations of relative translational dynamics in the orbital plane of chaser spacecraft, the issue of unknown orbit elements is solved. And for the problem for unknown inertial tensor, we propose a novel robust estimator named interaction cubature Kalman filter(In CKF) to handle it. The novel filter consists of multiple concurrent CKFs interlacing with a maximum a posteriori(MAP) estimator. The initial estimations provided by the multiple CKFs are used in a Bayesian framework to form description of posteriori probability about inertial tensor and the MAP estimator is applied to giving the optimal estimation. By exploiting special property of spherical-radial(SR) rule, a novel method with respect to approximating the likelihood probability of inertial tensor is presented. In addition, the issue about vision sensor's location inconformity with center mass of chaser spacecraft is also considered. The performance of this filter is demonstrated by the estimation problem of RVD at the final phase. And the simulation results show that the performance of In CKF is better than that of extended Kalman filter(EKF) and the estimation accuracy of pose and attitude is relatively high even in the completely non-cooperative case.
在系统分析与设计过程中,针对高阶动态系统所具有的时滞性,常常利用具有延迟环节的一阶(first order plus time delay,FOPTD)或者二阶系统(second order plus time delay,SOPTD)模型对其进行近似处理,由于建模误差过大影响所描述系统的准确性和控制性能。本文给出了具有延迟环节的新型非整数阶类一阶系统模型(non-integer order plus time delay,NIOPTD),并分别设计了某高阶系统降阶得到的传统模型与新型类一阶系统近似模型,对比分析新型类一阶系统模型的优点与可行性。针对上述3种系统模型(FOPTD、SOPTD、NIOPTD)在频域内给出分数阶PIλDμ控制器新的参数整定方法,通过仿真对比分析得出方法的有效性,并证实分数阶PIλDμ控制器作用于NIOPTD模型具有最好的控制性能和鲁棒稳定性。
