Automation and Remote Control最新文献

This paper considers nonlinear continuous-discrete (hybrid) systems containing two subsystems of differential and difference equations, respectively, and one-dimensional (scalar) or multidimensional (vector) control. The transition from a nonlinear hybrid system with a constant sampling step h > 0 to an equivalent, in a natural sense, nonlinear discrete dynamic system is presented. Sufficient conditions are established, first, for reducing the first approximation systems of nonlinear discrete systems to the Brunovský canonical form and, second, for stabilizing such systems and nonlinear hybrid systems with control of different dimensions. Algorithms for constructing stabilizing control laws for nonlinear hybrid systems are developed. Numerical examples are provided to illustrate the effectiveness of this approach to stabilizing nonlinear hybrid dynamic systems.

A smooth autonomous system of general form is considered. A global family of non-degenerate periodic solutions by the parameter h is constructed; the period varies monotonically on this family. The problem of stabilizing the oscillations of the reduced controlled system is solved. A smooth autonomous control law with a parameter depending on h is applied, and an attracting cycle is constructed. The results are concretized for an nth-order differential equation. The relation of these results with the conclusions obtained for the reversible mechanical system is established. An adaptive control scheme for the reduced conservative system is proposed to stabilize any oscillation of the family. Some applications are presented.

The problem of output-feedback compensation of bounded additive perturbations affecting a minimum-phase linear system with unknown parameters is considered. An adaptive auxiliary loop is developed, which does not require to know the perturbation model and allows one to: (a) separate the processes of estimation of parametric and additive perturbations, (b) estimate and compensate for the additive perturbation with any given accuracy if the conditions of the parametric identifiability are met. The above-mentioned separated estimation of two disturbances of different nature is achieved by augmentation of the A.M. Tsykunov auxiliary loop method with the law to identify the unknown parameters, which is based on the instrumental variables approach and the procedure of dynamic regressor extension and mixing (DREM). The obtained system of the additive perturbations compensation has a certain potential to be used together with the conventional industrial PI-, PID-controllers. The theoretical results of this study are validated via mathematical modelling.

This paper considers an oligopoly model with an arbitrary number of Cournot-reflexive agents under incomplete information in the classical case of linear cost and demand functions. Agents make decisions based on a collective behavior model. The problem of identifying convergence conditions to the model equilibrium is studied, with an emphasis placed on the trajectory of the sum of the action residuals of all agents. Aggregate estimates of this trajectory are obtained. They can be used to judge how the trajectory of each agent evolves towards the equilibrium.

The article introduces the concept of Marked Markov processes, which are used to study a repairable hot double redundant system with a single repair facility. It is assumed that components’ life- and repair times follow arbitrary distributions. The proposed approach allows for calculating the system’s reliability function, and its mean lifetime, as well as conducting the sensitivity analysis to the shape of input distributions. The new method was validated with numerical examples by comparing it with previously obtained analytical results and showed high accuracy.

Stability of regular precessions of a rigid body bounded by the ellipsoid of revolution in a free molecular flow of particles is studied. Conditions of stability of regular precessions of the body are obtained and the Poincare–Chetaev bifurcation diagrams are constructed.

In this paper, the disclosure of information about the scoring model is investigated. Some of the company’s customers find out their internal rating in the company. Such customers can change their behavior to increase their internal rating. The customers who are aware of the leakage are represented as players who can choose a strategy: whether to increase their internal rating and, if so, how much. The main goal is to find the Bayesian–Nash equilibrium in this game and find out how it depends on various parameters, such as the scale of the leakage, the distribution of ratings.

Identification of central elements in networks is an ill-defined problem. Hence, a large number of centrality measures have been proposed in the literature. We present a survey of existing axioms, which characterize certain properties of centralities. We also perform a perturbation analysis of centrality measures in real and artificial networks.

The paper considers a state observer-based iterative learning control design problem for discrete linear systems. To accelerate the convergence of the learning error, a combination of the heavy ball method from optimization theory and the vector Lyapunov function method for a class of two-dimensional systems known as repetitive processes is used to develop a new design. A supporting numerical example is given, including a comparison with an existing design.

This paper is devoted to a classical NP-hard problem, known as the three-index axial assignment problem. Within the corresponding framework, the problem of combining feasible solutions is posed as an assignment problem on the set of solutions containing only the components of selected feasible solutions. The issues of combining solutions for the multicriteria problem with different criteria convolutions are studied. In the general case, the combination problem turns out to be NP-hard. Polynomial solvability conditions are obtained for the combination problem.