Disturbance observer-based quadrotor attitude tracking control for aggressive maneuvers

In this paper, a disturbance observer-based quadrotor attitude controller for aggressive maneuvering is presented. The controller is made up of the cascade connection between two control-loops: an outer quaternion-based attitude control-loop and an inner disturbance observer-based angular velocity tracking control-loop. The disturbance observer is designed to estimate and compensate for the Coriolis term and the external disturbances. It is shown that, for fast maneuvers, the disturbance observer needs to take into account the motor dynamics. This allows to notably increase the observer bandwidth, leading to significant improvements in the disturbance rejection capabilities. The stability of the resulting closed-loop is analyzed. Also, different simulations and flight tests are carried out to validate the main results, showing an outstanding tracking performance when aggressive attitude maneuvers are being executed, even in the presence of strong disturbances such as suspended payloads.     

有限时间内异构多智能体系统的协同输出调节

针对异构多智能体系统有限时间输出调节问题,本文提出了分布式有限时间状态观测器控制算法和相应的输出反馈控制策略.应用图论、Lyapunov稳定性理论证明了所设计的观测器能够在有限时间内估计出外部系统状态,同时可以保证系统在有限时间内的稳定性,很好的解决了当部分跟随者无法获取外部命令时的有限时间输出调节问题.最后,通过MATLAB数值仿真验证了该协议的有效性和正确性,仿真结果表明系统除了在有限时间内获得更快的收敛速度外,还具有良好的暂态性能.     

线性/非线性切换扩张状态观测器

为了解决非线性扩张状态观测器(NLESO)对大幅度扰动估计能力有限的问题,本文提出一种线性/非线性切换扩张状态观测器.首先分析了非线性扩张状态观测器对大幅度扰动估计能力有限的原因,然后提出在NLESO的非线性区间引入一段线性扩张状态观测器(LESO)弥补NLESO的缺陷,其次从理论上证明了提出的线性/非线性切换扩张状态观测器的收敛性.最后,通过数值仿真验证了提出的线性/非线性切换扩张状态观测器的可行性.     

基于干扰观测器的机械臂改进趋近律的滑模控制

针对机械臂滑模控制中存在抖振的问题以及对外界干扰较为敏感的特性,设计了一种基于干扰观测器的机械臂改进趋近律的滑模控制策略。该策略在指数趋近律中引入饱和函数,提出了一种改进趋近律;根据系统动力学模型,对系统的不确定性和外部干扰采用干扰观测器进行观测;在此基础上设计系统滑模控制律,控制律对未观测的干扰进行补偿。仿真结果表明,控制策略能有效地抑制轨迹跟踪中的抖振问题,较好地克服外界干扰,保证系统的控制性能。     

A sliding mode observer design for uncertain fractional Ito stochastic systems with state delay

This paper addresses the robust observer design problem for uncertain time-delay fractional Ito stochastic systems. A sliding surface is proposed and it is shown that state estimates converge towards it and remain there for the subsequent time. Additionally, by constructing a novel Lyapunov functional, a sufficient condition for the stability of the sliding motion of the estimated states is given in the form of Linear Matrix Inequalities (LMIs). It is demonstrated that the state estimates are stabilizable in probability provided that the LMI is feasible. Moreover, a finite-time sliding mode control law based on the estimated states is proposed. Simulation examples are given to show the validity and effectiveness of the results.     

Almost Disturbance Decoupling for a Class of Nonlinear Systems Subject to Time‐Delays Via Sampled‐Data Output Feedback Control

This paper considers the problem of almost disturbance decoupling (ADD) via sampled‐data output feedback control for a class of uncertain nonlinear systems subject to time‐delays. Based on output feedback domination approach, a sampled‐data output feedback controller is designed to globally stabilize the system under a lower‐triangular linear growth condition. Gronwall‐Bellman‐like inequality and inductive method are introduced to estimate the state growth in the presence of time‐delays, uncertain nonlinearities and unknown disturbances. The proposed controller can attenuate the influence of disturbances on the output to an arbitrary degree in the L₂ gain sense. Finally, simulation results show the effectiveness of the control method.     

Design of Optimal Petri Net Supervisors for Flexible Manufacturing Systems via Weighted Inhibitor Arcs

This paper develops an approach to the design of an optimal Petri net supervisor that enforces liveness to flexible manufacturing systems. The supervisor contains a set of observer places with weighted inhibitor arcs. An observer place with a weighted inhibitor arc is used to forbid a net from yielding an illegal marking by inhibiting the firing of a transition at a marking while ensuring that all legal markings are preserved. A marking reduction technique is presented to decrease the number of considered markings, which can dramatically lower the computational burden of the proposed approach. An integer linear program is presented to simplify the supervisory structure by minimizing the number of observer places. Finally, several examples are used to shed light on the proposed approach which can lead to an optimal supervisor for the net models that cannot be optimally controlled via pure Petri net supervisors.     

Stabilization of Uncertain Multi‐Order Fractional Systems Based on the Extended State Observer

The extended state observer (ESO) based controller has been used successfully with integer‐order systems involving large uncertainties. In this paper, the robust control of uncertain multi‐order fractional‐order (FO) systems based on ESO is investigated. First, we transform the multi‐order FO system into an equivalent system in the form of a same‐order state‐space equation. Then, the ESO for the new system is established for estimating both the state and the total disturbance. Sufficient conditions for bounded‐input and bounded‐output stability are derived, and the asymptotic stability of the closed loop system is analyzed, based on whether the states are available or not. Finally, numerical simulations are presented to demonstrate the validity and feasibility of the proposed methodology.     

基于卷积神经网络的永磁同步电机转矩观测器

内嵌式永磁同步电机具有高功率密度、高可靠性和弱磁性等诸多优点,但由于电动机参数具有非线性化特征,导致电磁转矩难以精确估算。该文提出了一种基于卷积神经网络的电磁转矩估算方 法,即转矩观测器。首先,基于所搭建的高保真非线性内嵌式永磁同步电机模型,获得用于神经网络训练的转矩观测器数据;然后,基于所提出的卷积神经网络转矩观测器实现内嵌式永磁同步电机的精确控制;最后,为获取最优的转矩估算误差,在仿真实验阶段对不同参数和结构的卷积神经网络进行了对比和分析。结果表明,该神经网络可以实现电磁转矩的准确估算,所建立的转矩观测器具有良好的性能参数和泛化能力。