基于椭球束的多智能体系统执行器故障区间估计

Actuator Fault Interval Estimation for Multi-agent Systems Based on Ellipsoid Bundles

本文针对一类具有未知但有界不确定性的多智能系统,提出了一种基于椭球束的鲁棒故障区间估计方法。首先,将执行器故障向量视为辅助的状态向量,构建出一个新的增广系统。其次,基于图论理论,将多智能体系统整理成全局描述方式,针对该全局系统,利用全局相对输出估计误差,设计一种未知输入观测器,并生成估计误差动态系统。考虑估计误差动态系统受未知但有界的不确定性影响,基于椭球束生成估计误差每个时刻受干扰影响的上下界,再结合观测器对增广状态的点估计结果,生成状态与故障的估计区间。最后,基于多个飞机编队模型对本文所提方法进行了仿真验证。

This paper proposes a robust ellipsoid bundles based fault interval estimation method for multi-intelligent systems with unknown but bounded uncertainties.Firstly,the actuator fault vectors are regarded as auxiliary state vectors and a new augmented system is constructed.Secondly,based on the graph theory,the multi-agent systems are organized into a global description.For this global system,an observer is designed based on a global relative output estimation error vector.Considering that the estimation error dynamical system is subject to unknown but bounded uncertainty,an ellipsoid bundle-based approach generates upper and lower bounds on the estimation error at each moment,which are then combined with the results of the observer's point estimation of the augmented and broadened states to generate the estimation intervals of the states and faults.Finally,the method proposed in this paper is validated in simulation based on several aircraft models.