This paper studies mixed-integer nonlinear programs (MINLPs) in networked multi-agent systems, where agents determine their discrete and continuous decision variables to minimize the sum of local objective functions, subject to both local and global coupling constraints. The resulting large-scale constraint-coupled MINLPs (CC-MINLPs) are nonconvex and NP-hard. Our goal is to develop a distributed optimization algorithm to compute feasible solutions with guaranteed suboptimality bounds. Since the dual problem of CC-MINLPs can be solved using distributed convex optimization methods, the central challenge lies in the recovery of primal solutions. To address this challenge, we leverage the inherent convexification effect in CC-MINLPs to establish primal-dual near-optimality conditions. Based on these conditions, we propose a distributed algorithm for CC-MINLPs, which employs distributed projected subgradient methods to solve the dual problem and uses local running averages and minimization to recover primal solutions. We prove asymptotic primal feasibility and ensure finite-time feasibility by tightening the coupling constraints. Moreover, we characterize asymptotic and finite-time suboptimality bounds. Finally, numerical simulations validate the design and analysis.
This paper studies the bumpless transfercon trol of continuous-time switched linear systems with ad missible dual mode-dependent dwell time (ADMDT) switching signals. To reduce the inherent conservatism of existing dwell-time-independent bumpless control methods, the time-interpolation-based dual mode-dependent piecewise time-varying bumpless controller and multiple discontinu ousLyapunov function (MDLF) are designed bypartitioning the ADMDT into multiple time intervals. The overall performance of control system is improved as the number of segments increases. However, the discrete nature of MDLF conflicts with the continuous nature of bumpless controller, making the sufficient condition unsolvable. To this end, a new design approach is proposed in which the Lyapunov constraint parameters are embedded in the structures of MDLF and bumpless controller, achieving the coexistence of continuous inputs and MDLF. Thus, the proposed method naturally avoids the limitation of existing bumpless control methods, which require predesigned stabilizing controller gains to ensure conditional solvability. That is, the proposed method can actively balance the transition performance and stabilization performance, which alleviates the problem of system instability caused by input smoothing transition. Finally, the superiority and validity of the established theoretical results are illustrated by using a numerical example.
In this paper, we study remote state estimation of a discrete-time linear time-invariant (LTI) system over an additive white Gaussian noise (AWGN) channel with channel feedback, while ensuring that the state information remains confidential from an eavesdropper who intercepts the signal transmission via wiretap channels. Leveraging channel feedback, we propose an LTI encoder-decoder scheme and establish that secure remote state estimation is achievable if and only if the channel capacity exceeds the topological entropy of the plant. In contrast to prior results for the feedback absent case, where additional capacity is demanded to counteract the presence of wiretap channels, our result conveys an important message that feedback can be exploited to guarantee secrecy automatically without incurring additional capacity costs. A systematic design procedure for the encoder-decoder pair is also presented.
In this paper, we extend nonlinear negative imaginary (NI) systems theory to facilitate applications to power systems stability analysis and control in the presence of battery-based actuator saturation. We introduce a generalized definition of nonlinear NI systems that allows outputs with nonlinear input feedthrough, and we develop a Luré-Postnikov-type Lyapunov function to establish stability of positive-feedback interconnections between nonlinear NI plants and controllers. To address actuator saturation, we design an NI-preserving anti-windup compensator that guarantees the controller eventually exits the saturation region while maintaining closed-loop stability. The proposed framework provides an NI-based transient stability analysis method that complements the Equal Area Criterion by certifying a region of attraction for the after-fault equilibrium. We further design an angle-feedback controller that uses real-time measurements to coordinate large-scale battery actuation and enhance transient stability.
The paper considers a system of unidirectional first-order hyperbolic partial differential equations (PDEs) coupled with ordinary differential equations (ODEs), modeling two-phase advective transport processes subject to both boundary and distributed input delays. To address the system defined over two spatial domains, we design a set of three backstepping transformations involving both Volterra- and Fredholm-type integral operators, which introduce eight kernel functions: five associated with the PDE states and three with the ODE states. Since the distributed delayed variable spans the two spatial domains, its transformation must integrate over all state variables, leading to three-dimensional kernel functions with singular initial conditions involving Dirac delta functions that serve as sampling operators to establish mappings between variables of different dimensions. By employing the method of characteristics and successive approximations, we prove the well-posedness of the kernel equations. On this basis, we prove the invertibility of the Fredholm transformation and establish the exponential stability of the closed-loop system in the L^∞ norm under the delay-compensated controller. Simulation results for a linear example and a nonlinear screw extrusion model are presented to demonstrate the effectiveness of the proposed method.
