Modern data centers (DCs) face the problem of high energy consumption, which leads
to increased operating costs and negative environmental impact. This raises the need to develop
intelligent control systems (ICS) that can improve energy efficiency and optimize the use of
engineering infrastructure.
The aim of this work is to evaluate the effectiveness of integrating neural networks and a robust
PID controller into an IMS to improve control adaptability, reduce energy consumption and
increase the resilience of data centers to changing operating conditions.
Methods: in order to achieve the set objectives, a simulation model of the TIER IV level data
center was developed in the TRNSYS environment, which allows to simulate dynamic
processes of energy consumption. The model uses machine learning algorithms to predict
thermal load and power consumption, implemented using a neural network trained in MATLAB
environment. A robust PID controller is also implemented to control cooling systems based on
the predicted data. An economic analysis of the efficiency of IMS implementation with the
calculation
 of
 key
 indicators:
 ROI,
 NPV
 and
 BCR.
The novelty of the work consists in proposing an approach to balancing the input data for
neural networks, which allows to reduce the spread of amplitudes of oscillations, reduce the
probability of overtraining and increase the generalization ability of the network. For the first
time, a method of integrating neural networks and robust PID controller for data center energy
management, taking into account dynamic infrastructure changes, has been developed.
Result: The proposed control system reduced the PUE by 3.5%, reduced the power
consumption of the cooling systems to 40% of the total power consumption of the data center
and improved the accuracy of heat load forecasting. This increased the adaptability of the
control and provided a reduction in operating costs.
Practical significance: the developed methods and models are applicable for modernization of
data center engineering infrastructure in order to improve their energy efficiency, reduce costs
and ensure environmental sustainability. The results show the feasibility of implementing IMS
to improve the resilience and adaptability of data centers to changing operating conditions,
which helps to reduce operating costs and environmental impact.

The problem of detecting and recognizing road signs in difficult weather conditions is considered. An algorithm based on a combination of neural network models and possessing high accuracy and stability in recognizing images obtained in difficult weather conditions is proposed. The effectiveness of the algorithm was tested on a set of images overlapped by rain and snow. The experimental results showed a significant increase in the efficiency of the proposed algorithm compared to the algorithm that does not take into account the influence of the weather.

The article describes a methodology for finding optimal copies of high-frequency sound emitters based on objective measurements. Recommendations for automation of the selection process are given. The optimization problem of selecting serial copies or models of high-frequency emitters of sound was performed, a mathematical model of the selection algorithm was compiled.

Modern global navigation satellite systems in medium-altitude and geostationary orbits have two significant drawbacks. The first is the degradation of the availability level (according to the criterion of the geometric factor by location (PDOP) ≤ 6) and the average value of the geometric factor with an increase in the angle of reception of navigation signals due to natural or anthropogenic terrain, which leads to both a decrease in the accuracy of navigation definitions and a complete failure of navigation. The second is the low level of the navigation signal in the zone of use (the Earth's surface and atmosphere), which, with a significant level of radio interference, leads to the impossibility of navigation definitions.

The problems of classification of linguistic aggression in unstructured texts of the information space are considered. Various approaches to defining the concept of "linguistic aggression" are analyzed, as well as methods for its detection and assessment. Particular attention is paid to issues of automatic text processing for detecting aggressive statements. Existing algorithms and machine learning models used to solve this problem are described. In addition, the prospects for the development of text analysis technologies and their application in the fight against the spread of verbal aggression on the Internet are discussed.

The subject of the study. Results of the electromagnetic field strength forecasting in the very low frequency band highly depends on the accuracy of the input data, such as the underlying surface electrical characteristics, the characteristics of the geomagnetic field and the ionosphere. Currently, modern geophysical models provide sufficiently high accuracy of the initial data for the field strength forecasting, but their implementation is complicated by the disparity of data presentation formats and requirement for software implementations of forecasting methods refinement. The study proposes a comprehensive methodology for using modern geophysical models for profiling geophysical characteristics on the very low frequency band radio path.

Materials and methods. The complex technique was obtained by combining a number of partial techniques and implemented as a program in the Matlab programming language. The initial data are taken from digital global maps of the underlying surface electrical characteristics, geomagnetic models, such as International Geomagnetic Reference Field or World Magnetic Model, and the International Reference Ionosphere model. The technique is based on a cyclic access to the geophysical models which is controlled by the number of parameters such as the Sun zenith angle, the underlying surface conductivity, the geomagnetic azimuth of the radio path and the geomagnetic field amplitude. When the obtained value of one of the control parameters goes beyond the designated interval, the characteristics of the radio path segment are added to the initial data array.

Results. Approved software package for the field strength forecasting has shown that the replacement of the initial data models introduces visible periodic disturbances into the forecast, that take a form of the field strength level pulsation over a distance. The effect was quantified with the corresponding harmonics levels in the periodogram. The analysis showed that periodic disturbances are the result of a sharp decrease in electron density in the International Reference Ionosphere model, therefore, a hybrid ionospheric model has been introduced into the complex segmentation technique, that combines the bi-exponential and International Reference Ionosphere models for a smoother decrease in electron concentration at low altitudes.

Discussion and conclusions. Field strength forecasting with the complex technique revealed that the introduction of a hybrid model gives the least periodic disturbances values of the field strength. However, the use of a hybrid ionosphere model is complicated by the need for a rational selection of the point where takes place the transition from one exponential function to another, therefore, the use of a bi-exponential model requires additional research. Moreover, the further direction of the study is to assess the accuracy of the developed methodology forecasting based on practical measurement data.

Multiple input – multiple output (MIMO) technology, thanks to the beamforming and beam direction control (beamforming) system, is a key technology that has a significant impact on improving the spectral efficiency of fifth-generation (5G) wireless communication systems. The main idea of this technology is spatial separation of users, simultaneous transmission of symbols of several users on one element of the time-frequency resource due to their spatial multiplexing. Due to the possibility of repeated use of the same time-frequency resources in different beams of a common antenna system, the efficiency of using the frequency band increases. However, the widespread use of the same time-frequency resources in neighboring beams leads to interference between neighboring beams (interbeam interference) in the antenna system. Estimating the level of this interference is one of the most important procedures in spatial planning of modern wireless networks. This article proposes a technique for determining the influence of the direction and width of neighboring beams on the level of interference between these beams, which makes it possible to reduce the level of such interference by dividing the beam pattern into fractions. A new method of arranging adjacent beams is also proposed, which makes it possible to reduce the level of interference between the beams. Based on the conducted research, the gain from the proposed technique in the transmission rate in the beam has been determined. 

The paper describes a general approach to finding ways to resolve conflict situations with discrete mismatch using a mathematical apparatus. Within the framework of mathematical modeling of conflicts, a general method for constructing a conflict function for each subject participating in the conflict has been developed. The main tasks of conflict resolution have been identified. Using the example of two subjects participating in the conflict, the conditions have been obtained that link the dissatisfaction functions of the subjects, under which the considered problems of conflict resolution will have a solution. The application of the Monte Carlo method to the search for a solution to conflict situations with discrete disagreement for each conflict resolution task has been shown. The considered approach to constructing a dissatisfaction function described using individual psychological characteristics of a person involved in a conflict allows one to better understand the causes of a conflict and make the right decision on resolving a conflict situation. The identified main problems of conflict resolution from the point of view of mathematical modeling of conflicts and the given diagram for constructing the objective function for these tasks can be applied to any type of conflict of any complexity.

Conclusions are formulated about the applicability limits of previously obtained empirical relationships that allow analytical calculation of the average queue length estimate in a queuing system with an incoming flow in the form of a fractal shot process (FSNDP). Brief overviews of network traffic modeling in client-server systems based on the FSNDP process, the structure of the process itself, and the approach used to obtain the analytical relationships are also given.

The article considers an approach to solving the problem of classifying rumors in the
news based on production rules. Unverified information appearing on news sites has the nature
of information garbage and is capable of causing significant harm to consumers in some cases.
The problem being solved is non-trivial, relevant and has no standard solution.

In recent years, unmanned aircraft have been receiving active use. In a number of applications, first-person control of unmanned aircraft is used. First-person control requires the transmission of a video stream from an unmanned aircraft to an external pilot station. However, during network transmission, delay and frame loss of the video stream occurs. This results in
lower intensity and higher latency of the video stream on the monitor of the external pilot station. In the previous part of the research, the method of increasing the intensity of video stream was presented, its mathematical models were built. In related works: the possibility of reducing the time of first-person control desynchronization was researched; the structure of unmanned systems first person control information exchange system was presented. The second part of the present research summarizes the results obtained in earlier works and presents an adaptive system of video stream latency and intensity control. The system consists from three blocks: neural codec mode controller, information exchange links control unit and video intensity control unit in two modes: intensity enhancement and delay reduction.