ДОСЛІДЖЕННЯ РЕЗОНАНСНИХ ПЕРЕНАПРУГ У МАГІСТРАЛЬНИХ ЕЛЕКТРИЧНИХ МЕРЕЖАХ 750 кВ З НЕСИНУСОЇДАЛЬНИМ ДЖЕРЕЛОМ СПОТВОРЕННЯ ЗА ДОПОМОГОЮ ШТУЧНОЇ НЕЙРОННОЇ МЕРЕЖІ

• Datum: 2018
• 1. Verfasser: Кучанський, В.В.
• Format: Artikel
• Sprache: Ukrainian
• Veröffentlicht: Інститут електродинаміки Національної академії наук України 2018
• Schlagworte: резонансні перенапруги; несинусоїдальні режими; парні гармоніки; штучна нейронна мережа
• Online Zugang: https://prc.ied.org.ua/index.php/proceedings/article/view/193
• Назва журналу: Proceedings of the Institute of Electrodynamics of the National Academy of Sciences of Ukraine

Considers the possibility of using artificial neural networks for rapid decision-making in the event of prolonged overvoltages. The analysis of the specifics of the task of developing an express method for determining the characteristics of overvoltages and common methods for their solution through the use of artificial neural networks is carried out. The architecture of artificial multilayer neural networks, suitable for the realization of this task, has been applied. The resonant overvoltages arising from the connection of the autotransformer to the electrical network 750 kV are considered. The research was devoted to the actual scientific and practical task - the development of models for the analysis of resonance overvoltages. An artificial neural network of overvoltage control its debugging has been developed. The application of the developed network for the identification of factors that have the greatest influence on the appearance and multiplicity of overvoltages in electrical networks is explored. The presence of a large number of fuzzy factors that affect the accuracy of the determination of the characteristics of overvoltage data necessitated the use of an artificial neural network. The factors that influence the characteristics of abnormal overvoltages are revealed. The results of determination of overvoltage characteristics of such a class by artificial neural network are given. The results of determining the characteristics of overvoltages using an artificial neural network are given. In this paper, to solve the problem of determining the characteristics of overvoltages, neural network methods are considered that differ in their ability to establish nonlinear connections between the parameters of the extra-high voltage transmission line. To achieve this goal, the following tasks were formulated: to carry out the definitions of overvoltage characteristics by neural network methods; to build a model of the neural network, corresponding to the initial data of the transmission line; get the results of the forecast; to estimate the accuracy of the functioning of the constructed model. References 11, figures 5.