Pareto Optimization with Reverse Normal-Boundary Intersection for Power Plant Models

This paper presents the use of Pareto optimization techniques that was previously analyzed with small scale power plant models where results could be verified analytically. Further research has been conducted into the use of applying these techniques to large scale models and analyzing performance. These approaches have been verified to scale well to larger applications. Through the use of multi-objective particle swarm optimization and reverse normal-boundary intersection, reference governors can be developed for numerous types of power plant. Combining these two approaches allows decisions made from a Pareto front to be seamlessly mapped into a simple and fast single variable optimization process. These applications are demonstrated with a small oil-fired power plant model and a large-scale coal-fired power plant model.