The paper is devoted to the developing of the methodological framework for modeling of nonlinear dynamical objects. Such a framework allows performing the coordinate transformation and formulating the algorithm of its usage to transform the series mathematical models of the nonlinear dynamical objects into the interval piecewise-defined linear models and then translate them into parallel ones. The abovementioned algorithm is based on reducing domains size where the nonlinearities of studied objects exist. We offer to reduce these domains by defining intervals on which nonlinear functions can be bounded on the bottom and on the top by linear functions. Thus, we suggest replacing nonlinear functions with their top and bottom piecewise-linear boundaries which are determined by minimization of the above-mentioned domains' areas. Previously suggested boundaries are used to define the transfer functions of the considered objects but one can use them for the stability analysis and the synthesis of closed-loop systems with desired dynamic as well. Then, we apply partial fraction decomposition of the transfer functions to perform transformation of the mathematical models into parallel form. Usage of the proposed approach is one of the ways to decrease calculation time and to improve multi-core CPU usage while simulation of nonlinear dynamical objects is being performed as well as to simplify the study of transients and steady-state modes of considered objects and to construct closed-loop control systems for a wide range of the nonlinear dynamical objects.
The paper deals with the development of a control system for a mobile robot. Synthesis of this control system is performed with the usage of interval methods. Such an approach makes it possible to change the amplitude of voltage for DC drives, which work in sliding mode, during robot motion. One can use this approach to design robust control systems for control of different kind robots, which parameters are exactly not known or can be changed in known ranges. We suggest a multiloop structure for the control system and to split the problem of control mobile robot on several small problems, such as control of the electric drive, form the desired trajectory and defining control signal. Each abovementioned loop solves its own problem.
The paper deals with the development of scientific backgrounds for the first order infinite-dimensional chaotic systems’ design. We suggest performing this design by replacing real time with some virtual variable, which is defined as a function of time and a system state variable. This virtual variable can be considered as the "new" time and it can be used to define transformation operator, which allows us to describe system dynamics in the "new" time domain by using known differential equations of the chaotic systems. We use the Mackey-Glass system as the initial chaotic system and show two transformations for this system by using different transformation operators. These transformations gave us the possibility to define the high-frequency dynamics for two chaotic systems. These systems can be used in different secured applications.
The paper is devoted to the presentation of one of the approaches to nonlinear systems control. Such an approach allows using well-known methods of linear control theory to synthesize robust controller for nonlinear objects. The above-mentioned approach is based on interval approximation of a nonlinear function by piecewise linear functions. The above-mentioned functions simplify the mathematical model of control object and allow us to take into account possible parameters drift and unmeasured external disturbance effect. We offer to reduce domains where these functions are defined by performing the minimization of some energy cost function. This minimization allows us to decrease energy consumption from a power source and increase live tine of the autonomous object. Then, we use an interval model to solve an inverse dynamic problem and define a control algorithm, which is based on using sliding mode control methods. Usage of the closed-loop control systems, which are developed in the proposed way, allows us to form the transient and steady state of control object with desired characteristics.
Image denoising is an urgent task that arises in image processing and transmission. Images are an integral part of content transmitted in infocommunication systems. Traditional statistical image filtering algorithms are not always effective for the random nature of the noise spectrum. Image denoising by convolutional neural networks is a modern and effective approach. The article is devoted to demonstrate the possibilities of using denoising convolutional neural networks to solve one of the most difficult tasks that developers face when performing the transfer of graphical information in infocommunication systems - denoising. Unlike traditional algorithms, denoising convolutional neural networks have architecture features that allow them to perform effectively image filtering with unknown noise level. It is suggested to use denoising convolutional neural networks to generate a correction signal in the infocommunication system, which transmits a noisy image. The article proposes to use denoising convolutional neural networks to generate a correction signal in the infocommunication system, which transmits a noisy image. Such pre-trained correction elements on a large and diverse image dataset can be easily fine-tuned for special filtering tasks.
The N – i switching method made it possible to establish analytical relations between the constraints of the state coordinates of the electric drive and the parameters of sliding mode control systems. This problem is solved for the cases of optimization in time domain and modal control. This paper presents a computational algorithm that implements a parametric synthesis of a control system taking into account changes in the levels of constraints caused by the performance of various shapes transient trajectories.