The concept of building a fully optical on-board information and computing network with intelligent control

Abstract

In this paper, the analysis of existing approaches to the construction of fault-tolerant on-board systems at the hardware and software levels is carried out. The main advantages of building a BIVS based on fiber optics components are considered, in particular, for deploying a unified intelligent system (UIS) based on it, producing intelligent output in case of emergency situations. Scenarios have been developed to fend off special flight situations of aircraft that arise in flight as a result of the impact of an adverse factor or a combination of adverse factors and leading to a decrease in flight safety. Algorithms have been developed for the reconfiguration of the BIVS in case of emergency situations, both during flight and on the ground. The principles of forming a specialized knowledge base of information support (SBSIP) based on operational documentation are proposed. In order to simplify the formalization of knowledge stored in the SBZIP, as well as to simplify the modification and addition of the SBZIP when new data appears, logical programming methods are used. The methods of calculating the characteristics of BIVS configurations from SBZIP are also proposed. Methods of dynamic synthesis of new BIVS configurations for emergency situations arising during aircraft operation have been developed. At the same time, new configurations are understood to be configurations that are not initially present in the SBIP, but whose structure and composition are based on all possible incomplete solutions. To solve the synthesis problem, information from the state space is used, formalized using logical programming methods. Based on it, a complete graph of the system is constructed, taking into account its redundancy, and a search and synthesis of the most suitable BIVS architecture is performed, allowing to fend off the failures that have arisen. The method of branches and boundaries is used as a graph search method. Also, for this purpose, in the case of using a hardware accelerator, a neural network approach can be used, which consists in solving an optimization problem using a recurrent artificial neural network.