Reverse Normal-Boundary Intersection for Multi-Objective Optimization for Power Plant Operation

This paper examines the Pareto front arising in the multi-objective optimization problem for power plant operation. However, having the Pareto front alone only provides more information about the operation of the power plant. To utilize this information, first, a point from the Pareto front must be chosen, which for this application turns out to be a very simple choice. Then a method is required to quantitatively distinguish this point from the other points on the Pareto front so that a computer algorithm can consistently choose this point when optimizing power plant operation. This is accomplished by a single optimization problem inspired by the normal-boundary intersection approach for generating Pareto fronts. This in turn, creates a way to choose the correct point over and over again, and in addition, removes the need to use aggregation methods to quantify the different objectives. This creates a much more intuitive approach to mutli-objective optimization which will be referred to as reverse normal-boundary intersection.