Analytical models of decision-making theory in digital transport logistics

Abstract

The active introduction of information and communication (digital) technologies into the modern reality of transport systems operation requires the development of modeling methods for the creation of software that allows to do the following: determine the formal efficiency of decisions made in digital transport systems (DTS); process large volumes of DTS data; perform analysis of DTS functioning environment using artificial intelligence algorithms, analytical analogues of neural networks, etc. A digital transport system, being a complex system, is determined by a large set of formalized indicators (database) and requires finding effective solutions for a sufficiently large number of criteria or signs of effectiveness. Obtaining reliable solutions in multi-criteria information situations causes difficulties, which are objective. In most cases modern mathematical models artificially reduce multi-criteria information situations to single-criteria categories. This approach, based on the use of integral criteria, has a fundamental disadvantage - the use of integral criteria to obtain estimates of the efficiency of actions or processes in complex systems is characterized by a high level of subjectivism. In this case, the obtained solution may be acceptable, but not an objective result. Therefore, it is necessary to develop mathematical models for solving multi-criteria problems applicable to the solution of problems in complex transport systems, allowing one to operate in the environment of large databases for operational reconfiguration of a management system in conditions of uncertainty and/or possible counteraction of the external environment. The article presents the results of development of mathematical methods of modeling, which allow us to construct algorithms for solving optimization tasks, formulated as multi-criteria models and in the presence of a high degree of uncertainty in the interaction of the system with the environment.