Optimization of Multi-Criteria Solutions to Selection Problems

Abstract

The paper proposes an approach to forming solutions to multi-criteria selection problems from the standpoint of their ordering by a number of agreed criteria. Existing examples of solving such problems in some cases are on the based use of a number of assumptions that are not in practice fulfilled. The paper proposes to optimize solutions to multi-criteria selection problems by forming a Pareto-optimal subset of solutions with ordering of its elements by the level of approximation to the constructed virtual variant with the best values of the selected criteria and the amount of resources used. Comparison of the candidates for selection is on comparisons of values according to agreed criteria based, the definition of which is by a set of mathematical models that reproduce the dependences of the estimates of the values of the selected criteria on the attributes of the candidates provided. An example of optimizing the construction of a virtual data processing system is using virtual computers from providers of cloud processing services and technologies presented. The results obtained indicate the possibility of using the proposed approach as part of decision support subsystems to solve the problems of operational management of dynamic service and production processes.