Scanning the Issue*

2017 IEEE Transactions on Automatic Control  
A systematic design methodology for state observers for a large class of nonlinear systems with bounded exogenous inputs is proposed. System nonlinearities are characterized by an incremental quadratic constraint parametrized by a set of multiplier matrices. LMIs are developed to construct observer gains ensuring that a performance output based on the state estimation error satisfies a prescribed degree of accuracy. Furthermore, conditions guaranteeing estimation of the unknown inputs are
more » ... ed. The proposed scheme is illustrated through an example which does not satisfy the so-called "matching conditions." This paper discusses fixed-time synchronization of a class of complex networks with synchronizing and desynchronizing impulse in a unified way. Several sufficient conditions in terms of matrix inequalities are given to ensure that all the subsystems in the network are synchronized with an isolated system in a settling time, which is independent of the initial values of the systems. The key approaches are to exploit time-varying Lyapunov functions, impulse intervals, and convex combination techniques. The designed controller is continuous and includes no sign functions, which eliminates chattering phenomena. Conventional projection-based decentralized algorithms are incapable of handling high-dimensional constrained problems due to the complex projection step required. To address this problem, the authors propose a projection-free approach based on the Frank-Wolfe (FW) algorithm and the decentralized FW (DeFW) algorithm. The DeFW algorithm can be viewed as an inexact variant of FW algorithm. The authors show the convergence of DeFW for convex and nonconvex problems and provide a rate analysis. A consensus-based DeFW algorithm is described. The performance of the algorithm is tested on low-complexity robust matrix completion and sparse learning. This paper deals with distributed learning in a network of agents. The authors propose a distributed algorithm leading to consensus on a hypothesis that best explains a set of observations of some conditionally independent random processes. Consistency for the concentration of the beliefs around the optimal hypotheses set is established along with a nonasymptotic, explicit, and geometric convergence rate. An improved learning protocol with better scalability with respect to the number of nodes in the network is discussed in the case of a static network. In this paper, the authors consider an approximation to the chemical master equation known as the linear noise approximation. The authors derive structure-preserving model reduction algorithms of the linear noise approximation based on structured projections. The algorithms are shown to apply to a broad class of systems and to locally satisfy error bounds on the reduced approximation. Algorithms to compute these projections are then applied to a model of the yeast glycolysis pathway. An adaptive optimal control approach applicable to a wide class of large-scale, nonlinear systems is presented in this paper. The proposed approach avoids the so-called loss-of-stabilizability problem as well as the problem of poor transient. Moreover, it does not require the system model to be in a certain parametrized form and it is able to efficiently handle systems of large dimensions. Theoretical analysis establishes that the proposed methodology guarantees stability and exponential convergence to state trajectories that can be made as close as desired to the optimal ones. Recently, collective behaviors of multiagent systems over signed graphs have attracted the attention of the control community, and bipartite consensus is one of the topics. This paper extends bipartite consensus further to multiparty consensus over linear heterogeneous agents, and establishes the equivalence between the multiparty output consensus and the cooperative output consensus. A sufficient condition is presented for leader-follower multiparty output consensus of linear heterogeneous agents. The idea of multiparty consensus is applied to provide a solution to the problem of formation control with time-variant constraints. This paper unifies smoothing and interpolation for continuous-time linear stochastic systems. The authors derive balanced Mayne-Fraser-0018-9286
doi:10.1109/tac.2017.2758058 fatcat:menj6pkrtbh45gyglgh3qrgspy