This paper proposed an adaptive super‐twisting (AST)‐based fractional‐order sliding mode control algorithm under disturbances for redundantly actuated cable driving parallel robots. A novel fractional‐order non‐singular fast terminal sliding mode surface is constructed, and the fast response convergence and accuracy tracking performance can be obtained in the sliding mode phase; meanwhile, an auxiliary system is designed to overcome the adverse effects of the input saturation. Then, to compensate for the model uncertainty and external disturbances, the uncertainty and disturbance estimation is designed, and a novel AST algorithm is developed to suppress the estimation error without the known bound of the adaptive gains and guarantee the finite‐time convergence in the reaching phase. The Lyapunov stability theory is employed to prove the convergence of the proposed controller. Finally, numerical simulations are conducted in the commercial software of MATLAB/SIMULINK to demonstrate the effectiveness of the proposed controller.

Compared to control bandwidth, low‐frequency uncertainties or disturbances like step signals can be well rejected by many methods having two‐degree‐of‐freedom (2‐DOF). Due to robustness constraint, technically more challenging is the rejection of medium frequencies, especially for bandwidth‐limited systems. Here, the equivalent‐input‐disturbance (EID) approach is extended to deal with the main medium‐frequency oscillation of a pantograph‐catenary system. First, a general EID estimator is developed with a low‐frequency estimator as a special case. Then, a fair comparison is conducted to clarify the essential differences between the conventional 1‐DOF‐based and the developed 2‐DOF‐based control systems. Furthermore, a robust stability condition is derived for the 2‐DOF‐based closed‐loop control system. A design algorithm together with design guidelines is provided, where the frequency characteristics of the uncertainties are utilized in the parameter design. Finally, simulations are carried out to validate the developed 2‐DOF‐based method for the pantograph‐catenary system in realistic environment.

An asymptotic output feedback controller for the hydraulic systems with model uncertainty compensation and prescribed performance is proposed here, where state observers, funnel function and nonlinear adaptive controller are integrated via backstepping technology. Only position signal is available in the process of controller design, the velocity and load pressure are estimated by observers. And, the adaptive control law is carried out to approximate disturbance. Then, a funnel function is introduced to guarantee the prescribed tracking error performance and the potential singularity problem in prescribed performance control is avoided. Furthermore, the whole closed system is proved to be asymptotically stable using Lyapunov method. Comparative experimental results performed on hydraulic systems are studied to illustrate the effectiveness of the proposed controller.

This work proposes the use of a logarithmic quantizer to minimize the computational load on the onboard processor of missiles. The quantizer output discretizes the continuous guidance command to a finite set of predefined discrete levels. This in turn puts an end to the need for installing a powerful state of the art modern processor onboard a missile, while enabling designers to install a comparatively cost efficient and compact processor. To include robustness properties, the proposed guidance strategy adopts the artificial time‐delayed control (TDC) philosophy. The use of TDC methodology eliminates the conservative assumptions of a priori knowledge about uncertainty bounds as required by most state‐of‐the‐art robust control schemes. Thus, the proposed guidance law is able to achieve interception even in the presence of uncertainties while significantly reducing control updates due to the use of input quantizer. Input saturation is also considered for the proposed quantized time‐delay control (QTDC) based guidance strategy. The Uniformly Ultimately Bounded (UUB) convergence of the closed‐loop system states is demonstrated through the Lyapunov theory. Simulation studies involving various engagement scenarios and a comparative performance study of the QTDC guidance scheme with the conventional periodic time‐triggered TDC technique are provided to highlight the efficacy of the proposed approach. In this work, a quantized input time delay based control is used to significantly reduce the control updates while designing an efficient robust control strategy. A sense of time‐energy efficiency is introduced for the overall system and input saturation is also considered.

This article discusses the exponential synchronisation problem for multiple neural networks with time‐varying delay and a more general non‐Laplacian coupling matrix is considered. To reduce the redundant communication, two event‐based time‐varying delay hybrid impulsive control schemes are proposed with static event‐triggered condition and dynamic event‐triggered condition. These novel control schemes neatly combine the event‐triggered coupling strategy with the time‐varying delay hybrid impulsive control, where the impulsive instant is only determined by the specified event‐triggered condition. Moreover, a delay impulsive differential inequality is established to discuss the synchronisation problem for controlled multiple neural networks. By using Stolz's theorem and introducing the average event‐triggered delay impulsive gain, some less conservative sufficient conditions are derived to assure the stability of the time‐varying delay multiple neural networks. Furthermore, the Zeno behaviour can be eliminated effectively and the consumption of the computing resources can be reduced. Finally, some examples are given to verify the superiority of the results.

First, a static event‐triggered time‐varying delay hybrid impulsive control scheme was proposed to lessen the communication burden. Then, a time‐varying delay impulsive differential inequality was established to guarantee the synchronisation of multiple neural networks. Moreover, the authors also discussed the issue that continuously triggering and sampling can be excluded.