A multi-agent approach to optimal functioning indicators determining for the technological cycles of mining systems

Abstract

Research purpose is to determine the optimal production indicators of industrial enterprise technological cycles based on the application of the multi‑agent systems paradigm. Research methods: game theory methods, Markov decision model, methods of multi‑agent systems theory and methods of statistical analysis of production indicators. An application of multi-agent systems paradigm is one of the variants to improve a mining enterprise efficiency and to find a new approach for optimal production indicators determining. Then system components can be represented as intelligent agents acting on the base of utility functions, and the production process - as a multi‑agent system. Agents operate in some production environment and are able to perceive its states to make decisions about subsequent actions based on this information. They can play both a positive and a negative role in a mining enterprise production system from the position of continuity of technological cycle operation. The enterprise production system’s operation monitoring was carried out. Agents were grouped according to their influence onto operation continuity into two categories: “supporting continuity” and “causing incidents”, and the results were compiled into a final game matrix. The strategy of the “positive” industrial agent group was determined using the Wald criterion. It was the lower price of the game, which corresponded to the amount of potential damage. A risk matrix for continuity incidents was built on the base of calculated amount of potential damage, each agent’s contribution level at each stage of the technological cycle, and the overall effectiveness of the industrial agents. The main research resulted are in developing a formal model of the mining enterprise production system as a multi‑agent system; creating a model for assessing risks of production process continuity disruption due to the predominant influence of negative industrial agents in the system; identifying key performance parameters of technological cycles, including risks of continuity disruptions, time indicators, overall effectiveness agents.