Artificial Neural Network for Performance Prediction of Absorption Chillers of Large-Scale Solar-Air-Conditioning Systems.
Abstract
To satisfy the increasing cooling demand of buildings, solar-driven absorption machines can be used as an environmentally-friendly alternative to conventional vapor compression chillers. To improve the confidence in this technology, good simulation models and monitoring tools are needed to ensure optimal performance. As proposed by various authors, the application of artificial neural networks (ANN) proved useful for providing an easy-to-implement, yet very accurate way of predicting the performance of absorption machines. To further explore the limits of the ANN performance predictions, this paper attempts to apply the models to four of the currently largest solar thermal air-conditioning (SAC) plants, covering different system-layouts and climate conditions. This way, the ANN can be tested on operating large-scale systems and individual results for each SAC system can be analysed and compared.