Recognition of the importance of using artificial neural networks and genetic algorithms to optimize chiller operation

This paper presents the optimization of chillers operating using artificial neural networks and genetic algorithms. For the needs of generating chiller models, an artificial neural network was used, trained with data collected from an actual chiller. For that purpose the basic characteristics of artificial neural networks are shown as well as the process of making specific chiller models used for testing the results of application of the genetic algorithm in usage optimization. The optimal criteria with the shown steps for the use of the genetic algorithm and optimization results is also displayed in the paper. The results of use of artificial intelligence methods in optimization of chiller operation are verified through an actual office building model created in the simulation software EnergyPlus and through a series of experiments on an actual office building, equipped with a modern integrated BMS. (C) 2012 Elsevier B.V. All rights reserved.