Automation and Remote Control最新文献

This paper considers filtering for linear systems subjected to persistent exogenous disturbances. The filtering quality is characterized by the size of the bounding ellipsoid that contains the estimated output of the system. A regular approach is proposed to solve the nonfragile filtering problem. This problem consists in designing a filter matrix that withstands admissible variations of its coefficients. The concept of invariant ellipsoids is applied to reformulate the original problem in terms of linear matrix inequalities and reduce it to a parametric semidefinite programming problem easily solved numerically. This paper continues the series of author’s research works devoted to filtering under nonrandom bounded exogenous disturbances and measurement errors.

This paper demonstrates that robust control based on only a priori information about the object’s uncertainty can be significantly improved through the additional use of experimental data. Generalized H_∞-optimal controllers are designed for an unknown linear time-varying system on a finite horizon. These controllers optimize the damping level of exogenous and/or initial disturbances as well as the maximum deviation of the terminal state of the system. The design method does not require the persistent excitation condition or the rank condition, which ensure the identifiability of the system. As a result, the amount of experimental data can be significantly reduced.

This paper is dedicated to an optimization problem. Let A, B ⊂ ({{mathbb{R}}^{n}}) be compact convex sets. Consider the minimal number t⁰ > 0 such that t⁰B covers A after a shift to a vector x⁰ ∈ ({{mathbb{R}}^{n}}). The goal is to find t⁰ and x⁰. In the special case of B being a unit ball centered at zero, x⁰ and t⁰ are known as the Chebyshev center and the Chebyshev radius of A. This paper focuses on the case in which A and B are defined with their black-box support functions. An algorithm for solving such problems efficiently is suggested. The algorithm has a superlinear convergence rate, and it can solve hundred-dimensional test problems in a reasonable time, but some additional conditions on A and B are required to guarantee the presence of convergence. Additionally, the behavior of the algorithm for a simple special case is investigated, which leads to a number of theoretical results. Perturbations of this special case are also studied.

The time-optimal control problem for an arbitrary number of nonsynchronous oscillators with a limited scalar control is considered. An analytical investigation of the problem is performed. The property of strong accessibility and global controllability is proved, and a program control is found that brings the system from the origin to a fixed point in the shortest time. Trajectories satisfying both the motion equations of the system and the additional conditions based on the matrix nondegeneracy conditions of the relay control have been found for bringing a group of oscillators to the origin. Two classification methods of trajectories according to the number of control switchings are compared: the one based on the necessary extremum conditions and the Neustadt–Eaton numerical algorithm.

Coordination in multiagent systems with information influences is studied. In particular, a model of multiagent system in which information is transmitted with a constant delay for all agents is studied. Using the Nyquist criterion applied by Tsypkin for systems with delayed feedback, a formula is obtained for the boundary value of time-delay which is included as a parameter in the system of differential equations with an asymmetric constant Laplacian matrix. The condition of independence of stability from time-delay is founded. The results obtained generalize some previously results and can be applied in coordination analysis in a multiagent systems with complex protocol.

We consider a class of well-known high-order trinomial linear difference equations and analyze the non-asymptotic behavior of their solutions under non-zero initial conditions from the unit box. It is shown that, for certain subsets of coefficients in the stability domain, there always exist initial conditions leading to peak, a large deviation of solutions from the equilibrium position, and that these deviations may take arbitrarily large values. Various special cases are studied, numerical examples are presented.

The paper is devoted to the analysis of the feasibility domain of electric power systems. The problems of calculating feasible and marginal regimes of power systems, analyzing the geometry of the feasibility domain, and generating samples in this region are considered. Parallels are drawn with the works of B.T. Polyak on the analysis of the image of a quadratic map, modification of the Newton method and the development of methods for generating asymptotically uniform samples in areas with complex geometry. Particular attention is paid to Newton’s method with the transversality condition (TENR), its application for constructing a boundary oracle procedure and utilization for generating samples in the power system feasibility domain.

We consider some problems with a set-valued mapping, which can be reduced to minimization of a homogeneous Lipschitz function on the unit sphere. Latter problem can be solved in some cases with a first order algorithm—the gradient projection method. As one of the examples, the case when set-valued mapping is the reachable set of a linear autonomous controlled system is considered. In several settings, the linear convergence is proven. The methods used in proofs follow those introduced by B.T. Polyak for the case where Lezanski–Polyak–Lojasiewicz condition holds. Unlike algorithms that use approximation of the reachable set, the proposed algorithms depend far less on dimension and other parameters of the problem. Efficient error estimation is possible. Numerical experiments confirm the effectiveness of the considered approach. This approach can also be applied to various set-theoretical problems with general set-valued mappings.

The problem of finding the arrangement of closed-loop control system poles that minimizes an objective function is considered. The system optimality criterion is the value of the H_∞ norm of the frequency transfer function relative to the disturbance with constraints imposed on the system pole placement and the values of the H_∞ norm of the sensitivity function and the transfer function from measurement noise to control. An optimization problem is formulated as follows: the vector of variables consists of the characteristic polynomial roots of the closed loop system with the admissible values restricted to a given pole placement region; in addition to the optimality criterion, the objective function includes penalty elements for other constraints. It is proposed to use a logarithmic scale for the moduli of the characteristic polynomial roots as elements of the vector of variables. The multi-extremality problem of the objective function is solved using the multiple start procedure. A coordinate descent modification with a pair of coordinates varied simultaneously is used for search.

This paper is devoted to the problem of solving a system of nonlinear equations with an arbitrary but continuous vector function on the left-hand side. By assumption, the values of its components are the only a priori information available about this function. An approximate solution of the system is determined using some iterative method with parameters, and the qualitative properties of the method are assessed in terms of a quadratic residual functional. We propose a self-learning (reinforcement) procedure based on auxiliary Monte Carlo (MC) experiments, an exponential utility function, and a payoff function that implements Bellman’s optimality principle. A theorem on the strict monotonic decrease of the residual functional is proven.