ИДЕНТИФИКАЦИЯ КОСМИЧЕСКИХ ОБЪЕКТОВ НА ОСНОВЕ ИНТЕГРАЦИИ ДАННЫХ ОТ РАЗЛИЧНЫХ СИСТЕМ НАБЛЮДЕНИЯ И НЕЧЕТКОГО ВЫВОДА

Abstract

For monitoring near-earth space, one of the most important tasks of recognizing space objects, which includes subtasks of classification of space objects by type (spacecraft, launch vehicle, elements of launching or functioning of spacecraft, fragments of destruction, etc.) and its identification (nationality, intended purpose, degree of danger, functional state, etc.). The aim of the work is to solve the problem of increasing the efficiency and accuracy of various space objects based on the integration of data obtained from radar, radio engineering, optoelectronic and promising quantum-optical (laser-optical) means and processing them using algorithms of fuzzy inference and / or with using neural networks and fuzzy inference. Domestic and foreign means of monitoring near-earth space are considered, their technical characteristics and comparison are presented. The solution to this problem is justified by important national economic and environmental goals, since most of the space objects in the Earth's orbit are space debris. To solve the problem, a rule base is proposed for fuzzy conclusions of the most appropriate approach for determining various types of objects for given conditions and the composition of space control facilities. In addition, a fuzzy neural network was trained in the ANFIS editor using information and analytical reports from that multi-channel monitoring telescope MMT-9, the structure of the generated fuzzy neural network is shown. Based on the comparison, it is shown that the classification of space objects using neural networks and odd inference is more accurate than with fuzzy inference based on the Mamdani algorithm, but requires long training. It is shown that on the basis of increasing the efficiency of using the modern capabilities of space objects with high recognition accuracy. Conclusions are made about the results of using the use functions, numerical calculations and models in the Matlab environment are presented.