Aerospace Instrument-Making Annotation << Back THE PROBLEM OF USING NEURAL NETWORKS FOR MONITORING, DIAGNOSING AND PREDICTING THE TECHNICAL STATE OF AVIATION EQUIPMENT D.Y. Kiselev, D.O. Pushkarev The article analyzes the possibilities of using neural networks in aviation. Methods for using neural networks in the field of technical operation, automatic control systems, monitoring and diagnostics of aviation equipment are shown. It is shown that neural networks are a powerful tool that allows you to create mathematical models of an object and simulate various processes occurring in it and simulate random phenomena, thereby allowing you to detect the most complex dependencies. An analysis of neural networks was carried out and, as a result, it was found that neural networks are nonlinear in structure, which gives them an advantage over linear models that have been used for several decades. It is shown that the advantage of neural networks over linear models is the ability to use small samples to build complex dynamic models. The structure of neural networks and neurons is shown. Various options for activation functions of neural networks are considered. The advantages of neural networks are given, the main of which are self-learning, adaptability and scalability. Keywords: neural network; GTE; diagnostics; technical condition; expert system; fuzzy logic. DOI: 10.25791/aviakosmos.10.2020.1184 Pp. 34-41.