Applying Neural Network and Genetic Algorithm in Chiller System Optimization

The optimal use of electrical and fuel energy in a chiller system, like absorption chiller system, is important in achieving economical system operation. The related research work, including the derivation of a thermal model for the system by an integration of component models can be a tedious task, based on the technology today. The numerous assumptions associated with the governing equations and the nonlinear structure of the equation set as a result are often the limitations in computer simulation for producing reliable converging solutions. In many practical applications, like the selection of a global optimal supervisory control scheme, a system-based ANN modeling approach appears to be an attractive alternative. This paper describes the required process of deriving an artificial neural network model of a direct-fired double-effect absorption chiller system. Also discussed is a new concept of integrating neural network and genetic algorithm in the system optimal control.