|
|
 Âûïóñê 108
- Gorbunova A., Lebedev A. On a new approach to estimating response time quantiles of a fork-join queueing system
The article proposes a new approach to estimating the quantiles of the response time distribution of the fork-join queueing system. We consider a classic version of this system with a Poisson input flow and exponential service times on homogeneous servers. Upon receipt of tasks into the system, they are instantly divided into a fixed number of subtasks and sent for service to the appropriate subsystem with an unlimited capacity storage device and one server. The task is considered served after all its components have been serviced. This system allows you to simulate many real processes, which, in order to increase efficiency, are characterized by dividing large tasks into smaller components, for example, parallel or distributed computing systems. The difficulty of analyzing systems lies in the presence of a dependence between the sojourn times of subtasks, which significantly complicates the analysis of all performance characteristics of such systems. The main contribution of the article is the approach to determining the quantiles of the response time distribution, the assessment of which is no less valuable than the assessment of the mean response time. At the same time, a much larger number of works in this area are devoted to calculating the mean, which is explained, among other things, by the complexity of carrying out such an analysis even for a given characteristic, and estimating quantiles seems to be an even more laborious task.
- Nazarov A., Rozhkova S., Titarenko E. Asymptotic analysis of the M^[n]/GI/1 system with the remaining service time
À single server queuing system with Poisson batch incoming stream, repeated calls, instant and delayed feedbacks is considered. It is assumed that service time is distributed according to an arbitrary law, and the service durations are independent of each other. When the server is busy, incoming customers are sent into orbit. The problem is to investigate a random process of the number of customers in orbit. When compiling the Kolmogorov equations for the system, an additional variable is used - the remaining service time. The resulting system of equations is solved by the method of asymptotic analysis under the condition of a large delay of customers in orbit. As a result, a stationary probability distribution for the number of customers in orbit was found. The resulting asymptotic distribution is compared with the distribution found in previous papers for the case of an exponentially distributed service time. A numerical example is considered for a system in which the service duration has a gamma distribution with different parameters.
- Polin E., Moiseeva S., Moiseev A. Application of negative binomial distribution to approximate the stationary distribution of the number of arrivals in a QS with an incoming MAP, the intensity of which depends on the state of the system
This paper considers a mathematical model of an infinite-linear queuing system with an incoming MAP with an intensity depending on the number of busy servers. The parameters of the incoming process, namely its conditional intensities, change every time the state of the system changes, that is, a new request appears or one of the requests completes servicing. The service discipline is determined by the fact that the request occupies any of the free devices in the system on which its service is performed for a random time distributed according to an exponential distribution. For this model, obtaining a stationary probability distribution of the number of applications in the system by analytical means is not possible, so this paper proposes a heuristic approach, namely, the use of a negative binomial distribution as an approximation for the desired distribution. Two approaches to such approximation are proposed, for which a numerical analysis of the accuracy is performed based on comparison with the results of simulation modeling. The first approach is based on calculating the parameters of the negative binomial distribution using the exact values of the expected value and dispersion of the number of applications in the system under consideration, and the second is based on the fact that the intensity of incoming applications is determined by the Markov chain controlling the incoming process. It was found that the first approximation method gives more accurate results, however, when the system is heavily loaded, both approximations have a large error.
- Shumov V. Models and methods of justification density of boundary forces, providing reliable security state border
Based on the methods of management theory and the doctrine of border security, an approach has been formulated to substantiate the density of border forces that ensure reliable protection of state borders. Taking into account the provisions of military science and border statistics data, the required probabilities of detaining violators were assigned for three levels of security (high, medium, minimum). For economically motivated border violators, the justification for the required security density is made using models of border deterrence and the aggregated detention function. Using an example, the methodology for substantiating average security densities is considered. For border violators with non-economic motives (sabotage and terrorist groups), their deterrence is ensured by achieving the required probability of their detention (neutralization). In this case, the required security densities are calculated using the functions of border forces and means, analytical and simulation models. A simplified method for calculating the probabilities of apprehending violators is presented. The numerical values of the model parameters are estimated using a sufficient amount of border statistics data. A promising area of research is the formation of a promising image of border units.
- Parsegov S., Pugach M., Erofeeva V. Methods for calculating the state of charge of vanadium redox flow batteries: an analysis of the relationships
In the future, energy storage systems are expected to play a key role in the transition to low-carbon energy systems. Increasing renewable energy deployment rates require the integration of batteries that ensure the stability and security of the power system and mitigate the unstable behavior of renewables. Efficient operation of the batteries themselves is also important: it extends their life, reducing operating costs. Monitoring of the state of charge is one of the key tasks to help battery operation. This paper studies the existing explicit methods for calculation of the state of charge of vanadium flow batteries: the open-circuit voltage-based method and the Coulomb counter method. The relationship between them and the possibility of using them together to achieve more reliable and accurate charge state monitoring is investigated. In contrast to existing works, we derive an analytical expression for the overall state of charge that takes into account both main components of the battery, namely its stack and tanks. Analyzing their contributions reveals some shortcomings of existing approaches widely used to calculate and monitor the state of charge of flow batteries.
- Diane S., Vytovtov K., Barabanova E. Algorithm for analysis of multispectral aerial images from uav for identification of water pollution using analytical methods and neural network approaches
The article is devoted to the development of algorithms for the analysis of pollution on the surface of water bodies based on visual information obtained using a multispectral camera mounted on the body of a UAV. The structure of the algorithmic complex for the analysis of multispectral aerial photographs is proposed. Within the framework of the developed approach, each of the analyzed images undergoes a preprocessing procedure that ensures the alignment and alignment of its spectral channels into a single multidimensional raster. The developed analytical algorithm makes it possible to process and convolve the channels of a multispectral image using three mathematical operators - bandpass filtering, contrast change, and brightness change. At the same time, the choice of parameters for identifying pollution on the surface of water bodies is based on a preliminary stage associated with maximizing the contrast excess index for the reference area. The proposed neural network pollution analysis algorithm is based on the application of the sliding window method in combination with the convolutional architecture of the neural network classifier for the analysis of image fragments located on a rectangular grid. The software implementation of these algorithms, as well as the development of a graphical user interface, made it possible to confirm the assumption about the effectiveness of each of the considered approaches. Experimental studies have shown that the neural network algorithm wins in accuracy, and the analytical approach is easier to interpret from the point of view of an expert.
- Kozitsin I. Predicting public opinion dynamics with the SCARDO-model
Over the past 20 years, the theory of agent-based social influence models has been actively developing, a trend which is associated with the need to explain opinion formation processes in the context of the digitalization of communication channels and the intensification of information exchange processes. However, the practical side of this theory remains poorly studied. The key reason for this is the difficulties in calibrating model parameters and thus constructing an empirical foundation. This paper validates the SCARDO-model of opinion formation using empirical longitudinal data from the social network VKontakte. The data include three opinion snapshots of a large-scale sample of VKontakte users and a snapshot of their friendship connections. The model parameters are calibrated on the first two snapshots, whereas the third one is used to check the accuracy of the model’s forecast regarding the populations of opinion fractions at the next time moment. The constant trend model serves as a benchmark. The analysis performed shows that, depending on the method of parameter calibration, the prediction of the SCARDO-model can be more or less accurate than those of the constant trend model. At the same time, changes in public opinion in the dataset at hand (despite being sufficient to calibrate the model parameters) are small from the macro-scale point of view and, as a result, the typical value of the forecast error does not exceed one percent of "votes".
- Nizhegorodtsev R., Roslyakova N., Goridko N. Logistic models of the technology life cycle as a tool for assessing the efficiency of R&D expenditures for knowledge intensive companies
The study of the life cycles of technologies, their quantifications and the definition of breakpoints is an urgent scientific task. The most well-founded theoretical construction of the technology life cycle dynamics study is the logistic curve. The basis is a comparison of the dynamic series of costs and effects. The paper deals with the calculation of logistics trends expressing the relationship between annual data of gross revenue (effects) and R&D expenditures for Yandex in 2009-2021 (costs). Based on the approximation carried out by methods of nonlinear regression analysis, the values of maximum integral efficiency and maximum differential (point) efficiency of R&D expenditures for each of the considered time intervals are calculated. The study of logistics trends and the presented tools and results allow us to reveal the periods of dominance of one or another technological (or organizational and managerial) paradigm in the life of a certain high-tech company based on a comparison of aggregate and/or instantaneous efficiency for different periods of the company's development. In addition, the proposed results are relevant for assessing the prospects of technological shifts in the development of a high-tech company, namely, determining the level of technological or cost upper limit, expressed by the upper horizontal asymptote of the corresponding logistics.
- Schultz V., Chernov I. The utilization of virtual structures in the formation of scenario-cognitive models based on the utilization of expert knowledge
This study is dedicated to investigating the problem of enhancing the adequacy of scenario-cognitive models based on expert knowledge within a limited set of factors. One of the most important tasks in the formation of a scenario-cognitive model based on expert knowledge is the problem of taking into account the total influence of the external environment, i.e., those factors that remain outside the structure of the model, but influence the achievement of the required accuracy of modeling results. When constructing scenario-cognitive models of complex socio-economic and political systems, it is usually necessary to apply a significant simplification, which consists in concluding all the diversity of factors and connections between them in a relatively simple and understandable model. The quality of a model built on the basis of combining expert data should be determined by the adequacy of the image of a real object or situation. Consequently, when forming models using expert knowledge, it is also necessary to “expertly close” the structure of the model with some virtual substructures that are capable of generating certain signals reflecting the influence of the external environment. Typical signals simulating the influences of the external environment are presented. Typical structures of expert identification of the impact of the external environment on scenario model factors are introduced. An overall pattern of the scenario-cognitive model is presented, which is formed based on expert knowledge and consists of a multitude of actual factors of a complex system and proxy structures.
- Brokarev I., Vaskovskii S., Farhadov M. Automated information system for natural gas quality analysis
The article proposes an automated information system for determining the energy parameters of natural gas, shows its main components, and shows a version of the system’s operation based on experimental data. The proposed architecture of the automated system consists of the following parts: an information subsystem that implements the developed algorithms, a measurement information subsystem, and an analysis subsystem. The main advantages of the method and the system based on it are as follows: multifunctionality, which allows to modify the system at each stage for a specific task and obtain the necessary measurement information using commercially available and relatively inexpensive measuring equipment. Development of a model for solving the problem of analyzing the quality of natural gas includes a number of successive stages, namely: selection of data for training the model; choice of model architecture; choosing a model training method; assessing the accuracy of the model. The system was tested using the results of experiments conducted in laboratory conditions using data from real gas mixtures. Indicators of the accuracy of determining energy parameters have been calculated, from which it can be concluded that the method under study and the system based on it can be used to analyze the quality of real gas mixtures. The implemented architecture of the automated information system is information and computing, providing analysis of gas quality with low time costs.
- Romanova V., Zuev S. Adaptive trajectory control system AUV based on a direct propagation neural network
The paper is devoted to the development and study of an intelligent system of adaptive automatic control with a given target based on the use of artificial neural network of forward propagation. The control object is an autonomous unmanned underwater vehicle (AUV). In this paper, it is proposed to feed the signals received from the systems of the AUV to the input of the neural network, and use the output signal of the neural network for control to keep the vehicle on a given trajectory. As a result of this work, a model and a learning method are proposed that lead to holding the ANPA on a given trajectory under an external influence with a natural constraint for the considered mobile robot. Given a continuous preset trajectory and discrete signals from the ANPA systems, this allows following the preset trajectory with a simple intelligent control system that does not require large computational power. The proposed method of control system training allows pre-training on a numerical model of vehicle motion with random external influences, but does not require such pre-training under certain conditions. It is shown, in particular, that in the case of a sufficiently large learning rate, the model has time to rearrange itself and reacts to changed circumstances. The proposed intelligent system of adaptive automatic control can find application in those cases when the characteristic time of changes in the system is of the order of the training time, and the trajectory of motion satisfies the requirements stated in the paper.
|
|
