http://ea.donntu.ru/handle/123456789/29558 2013 г. Tue, 21 Apr 2026 08:55:42 GMT 2026-04-21T08:55:42Z http://ea.donntu.ru/handle/123456789/29698 Title: Синтез изображений рельефов и ландшафтов методом ROAM на кластерной и CUDA-архитектурах вычислительных систем Authors: Зори, С.А.; Лисеенко, В.В.; Зорі, С.А.; Лісеєнко, В.В.; Zori, S.A.; Liseienko, V.V. Abstract: В работе рассмотрена задача синтеза реалистичных изображений рельефов и ландшафтов и эффективность ее решения в реальном времени методом ROAM на популярных современных архитектурах параллельных вычислительных систем. Выполнено исследование эффективности реализации алгоритма ROAM на параллельном GPU-мультипроцессоре с использованием технологии CUDA и кластерной архитектуре NeClus с использованием технологии MPI. Показано, что использование GPU-мультипроцессора для реализации ROAM является неэффективным, а использование кластерной архитектуры – эффективным и позволяющим удовлетворить требования реального времени решением. Description: The paper considers the problem of relief and landscapes image-synthesis and the efficiency of real time relief image-synthesis by algorithm ROAM for popular modern parallel computing systems. The analysis and classifi-cation of the basic image-synthesis methods for reliefs and landscapes were made. We studied the efficiency of realization of ROAM-algorithm on parallel GPU with the use of CUDA-technology and cluster architecture NeClus with the use of MPI-technology and obtained some parameters of the synthesis process. The conclusion is that the use of GPU-multiprocessor for ROAM-realization is inefficient, while the use of cluster architecture is an effective solution, which meets the requirements of real time. Tue, 01 Jan 2013 00:00:00 GMT http://ea.donntu.ru/handle/123456789/29698 2013-01-01T00:00:00Z http://ea.donntu.ru/handle/123456789/29697 Title: Усовершенствование метаэвристических методов для решения задачи поиска оптимального маршрута Authors: Федоров, Е.Е.; Мартынова, О.П.; Слесорайтите, Э.; Федоров, Є.Є.; Мартинова, О.П.; Слесорайтите, Е.; Fedorov, Ye.Ye.; Martynova, O.P.; Sliesoraityte, E. Abstract: В данной работе рассматриваются такие метаэвристические методы поиска оптимального маршрута как генетический алгоритм, который основан на механизмах естественной эволюции, и алгоритм клонального отбора, основанный на особенностях иммунных систем. Для повышения вероятности выбора оптимального маршрута и уменьшения времени поиска в статье была осуществлена модификация операторов этих алгоритмов, которая использует имитацию отжига для учета начальных и конечных стадий работы алгоритмов. Для исследования разработанных модификаций в статье проведено численное исследование. Description: Now the city of Donetsk plays one of the key roles in transportation of various cargoes. It becomes the “air gate” between the West and the East, international automobile highways pass through this city, etc. In these conditions the important role is played by transport logistics. One of the main tasks of transport logistics is to find the optimum route between points. This task employs such artificial intelligence methods as Hopfield neural network, annealing imitation method, tabu search, ant algorithm, genetic algorithm, clonal selection algorithm, etc. The latter two have been improved in this work. The advantage of the offered genetic algorithm updating is that at early stages of its work equiprobable selection for the reproduction operator, outbreeding for the crossingover operator and the equiprobable scheme of the reduction operator are used. That ensures examination of the entire search space (random selection of chromosomes), and at final stages linearly ordered selection for the reproduction operator, inbreeding for the crossingover operator and the selection scheme of the reduction operator are used; it makes the search guided (current best chromosomes remain). Besides, the offered updating does not require scaling and can be used in case of fitness function minimization that is important for the minimum routesearch. The advantage of the offered updating of the clonal selection algorithm is that at early stages of its work all antibodies are cloned, new antibodies are added and the equiprobable scheme is applied. It ensures examination of the entire search space and at final stages only the best antibodies are cloned, new antibodies are not added and the selection scheme is used, which makes the search guided. The both updatings have increased the speed of search by 20%, and the probability the optimum route choice by 5%. The offered improved metaheuristics can be applied in automatic and automated computer systems for optimum route calculation when solving logistic and communication problems. Tue, 01 Jan 2013 00:00:00 GMT http://ea.donntu.ru/handle/123456789/29697 2013-01-01T00:00:00Z http://ea.donntu.ru/handle/123456789/29696 Title: Учет особенностей строения сетчатки глаза человека для выделения контуров объектов изображений Authors: Вовк, О.Л.; Vovk, O.L. Abstract: В данной работе подход к выделению контуров объектов изображений основан на принципах формирования зрительного образа в углублении сетчатки глаза человека. Предлагаемый метод моделирует возбуждение нейронов сетчатки на слоях фоторецепторов, горизонтальных, биполярных, амакриновых и ганглиозных клеток. Полученные результаты выделения контуров рассматриваемым методом сравниваются с результатами работы метода Собеля. Description: Recognition of image objects contours is an important task in many areas of our lives. Most often this problem is solved in the analysis of medical images, in flaw detection in the workplace, in reading graphic and symbolic information. In this paper a new method of image objects contours detection is described. This method is the modeling of human eye retina features. The considered approach consists in imitation of the basic layers of the retina’s central area. We consider the following layers: photoreceptors, horizontal cells, bi polar cells, amacrine cells and ganglion cells. It should be noted that the proposed model considers only the central area of the retina, where each photoreceptor has an almost straight connection with the ganglion cells. The method considers the features of horizontal cells, which have an inhibitory effect on bipolar cells. We analyzed two types of bipolar cells – those with an on-center and with an off-center. These types of cells correspond to illumination and darkening of the central area photoreceptors respectively. The input data for this method are the pixels of the original image, given with brightness characteristics. The transformations result in a black-and-white image, where white pixels correspond to the image object’s contour. Visual comparison of the results obtained using the proposed method and the Sobel method showed the advantage of the developed method for object contours detection with small changes in brightness. The presence of small noises within the contours should be mentioned as a drawback of the new method. Our further study will be focused on improving the method by providing closed contours and minimizing the noise in the contour. Tue, 01 Jan 2013 00:00:00 GMT http://ea.donntu.ru/handle/123456789/29696 2013-01-01T00:00:00Z http://ea.donntu.ru/handle/123456789/29695 Title: Поиск аналитического решения модельной задачи о напорной фильтрации газовой смеси в сплошной среде (на примере пневмообработки угольного пласта) Authors: Добровольский, Ю.Н.; Добровольський, Ю.М.; Dobrovolskiy, Yu.N. Abstract: Впервые предложены аналитические методы при решении задачи о напорной фильтрации газовой смеси в сплошной среде. Приведены теоретические сведения и практические аспекты применения методов. Description: Analytical methods are first proposed whensolving the problem about pressure filtration of gas mixture in continuum. Theoretical information and practical aspects of methods application are considered. A system of non-linear differential equations of parabolic type, which describes the process of pressure air forcing into a coal layer, is given. Five analytical solutions of the problem were obtained. It is supposed that external influence happens at a constant temperature T, or temperature T is set discretely a priori. Thus, it is necessary to find the solution of the first equation of the system, the last equation of the system is satisfied identically. Owing to uniformity of nonlinear differential operators in the other equations of the system, we can take the solution of the first equation of the system stretched or squeezed on the axis of functions as the solution for these equations. Method №1 is the simplest method of division of variables. The solution of the equation in the form of the product of functions of various arguments is searched for. Method №2 is searching the solution implicitly. A linear function relative to the main arguments is selected, and then a certain additive function from the solution is added. Method №3 is searching the solution in the form of square function relative to one argument. The method of undeter-mined coefficients is applied to determine the functions of the other argument. Method №4: a solution with generalized division of variables is obtained, and then differentiation method is applied. Method №5 is searching the solution in a parametrical form. A parameter depending on function’s arguments is entered. One argument of the function is recorded linearly relative to the entered parameter, and the function is recorded quadratically relative to the parameter. The function of the other argument is to be determined. Not all the obtained analytical solutions make clear physical sense, but can be used as “tests”for checking the correctness and accuracy of various numerical, asymptotic and approximate analytical methods. Solutions can form a basis for development of new numerical, asymptotic and approximate methods, which in their turn, allow studying more complex problems of heat and mass transfer, which do not have an exact analytical solution. Tue, 01 Jan 2013 00:00:00 GMT http://ea.donntu.ru/handle/123456789/29695 2013-01-01T00:00:00Z