APPLICATION OF MULTI-INDICATOR QUALITY ASSESSMENT OF PARETO APPROXIMATION IN MULTI-CRITERIA DECISION MAKING

Supporting multi-criteria decision-making based on multi-indicator quality assessment of Pareto approximation is an important task in the field of multi-criteria optimization. Such assessment allows making more informed and accurate decisions using information about several indicators that determine the quality of solutions. Pareto approximation is an approximate set of solutions that seeks to describe the real Pareto front as accurately as possible. In this case, it is necessary to assess how well the found solutions cover or approximate the real Pareto front. Among the large number of known algorithms for solving this problem, algorithms based on the preliminary construction of an approximation of its Pareto front (set) and called P-algorithms can be named. P-algorithms can be built on the basis of evolutionary and primarily on the basis of genetic algorithms, as well as on the basis of swarm algorithms of global optimization, such as particle swarm algorithms, ant colonies, honey bees, etc. Given the large number of P-algorithms, the problem of choosing the "best" algorithm for a given multi-criteria optimization problem (MCO problem) arises - the problem of meta-optimization. In this regard, a significant number of indicators of the effectiveness of P-algorithms (P-indicators) have been developed, which are based primarily on assessing the quality of the obtained approximation of the Pareto front (set) (P-approximation). Thus, the problem of assessing the quality of the P-algorithm itself becomes multi-criteria, or more precisely, multi-indicator.