Search

Currently, a large number of different mathematical models and methods aimed at solving problems of multidimensional optimization and modeling of complex behavioral systems have been developed. One of the areas of search for solutions is the search for solutions in conditions of incomplete information and the need to take into account changing external factors. Often such problems are solved by the method of complete search. In some conditions, the method of complete search can be significantly improved through the implementation and use of behavioral models of natural formations. Examples of such formations can be group behavior of insects, birds, fish, various flocks, etc. The idea of copying group activity of a shoal of fishes at the decision of problems of joint activity on extraction of food is used in work. The reasoning based on the simulation of the behavior of such a natural object allowed to justify the choice as a mathematical model - cellular automata. The paper examines the key features of such a model. Modeling of his work is carried out, strategies of behavior of group of mobile objects at search of the purposes are developed, key characteristics are investigated and the method of adaptive choice of strategy and change of rules of behavior taking into account features of the solved problem is developed. The search strategy is implemented in the work, which takes into account the need to solve the optimization problem on two parameters. The obtained results testify to the high descriptive possibility of such an approach, the possibility of finding the optimal strategy for the behavior of the cellular automaton and the formalization of the process of selecting the parameters of its operation. A further improvement of this approach can be the implementation of simulation to study the properties of the developed model, the formation of the optimal set of rules and parameters of the machine for the whole set of tasks.

The paper describes the structure of a method for forecasting of unfavorable aircraft accidents in the flight that uses convolutional neural networks (CNN) and recurrent neural networks (RNN) based on LSTM modules. In this work, forecasting means solving the problem of analyzing text messages. They are presented as а structured and unstructured text and are formed based on data obtained from various sources of information in the process of air traffic control. The procedures for determining hyperactive parameters and training a hybrid neural network model for forecasting of unfavorable aircraft accidents in the flight using CNN networks and LSTM modules are considered in detail.