Dolhorukov S. Computer-aided design of a navigation equipment test table

The aim of the research is to improve the efficiency of navigation equipment test tables (NETT), as well as its elements, by building a computer-aided design system based on the use of intelligent approaches, in particular reinforcement learning. A multi-agent approach for NETT design automation is proposed, which differs from the known ones in that it uses design agents with elements of artificial intelligence (AI) - reinforcement learning to solve a multi-criteria decision-making problem, resulting in improved design decision search in terms of using prior knowledge. The approach not only makes it easy to incorporate new features when needed, but also weighs their relative importance depending on the needs of the particular NETT design problem. A method for building an AI CAD system has been developed that differs from the known ones in that it integrates existing CAD systems into a single autonomous complex in which a human operator defines the criteria and constraints within which an AI-enabled multi-agent system is allowed to manage the automated operations, which leads to design time savings. For the first time, a method for solving the problem of multi-criteria design decision-making is proposed, which in its implementation uses agents with autonomous learning that independently collect data, generate new knowledge and use it to adjust the decision-making process, which allows direct optimization of target parameters without the need to define the model and form of approximation of these functions.