A T-S Model Based on Adaptive Fuzzy Neural Network for Liquid Desiccant Air Conditioning

Abstract

LDAC (liquid desiccant air conditioning) has been an attention for energy conservation for building environment management. To fulfil the requirement of independent air temperature and humidity control efficiently, it is essential to have an in-depth knowledge of dynamic characteristics for heat and mass transfer in the dehumidifier. In this paper, a modelling method based on fuzzy logical and adaptive neural network is presented to describe the dynamic characteristics of heat and mass transfer between air and desiccant solution in dehumidifier. Fuzzy logic and neural network are integrated effectively to make the proposed model with better dynamic response. Reliable large amount of experimental data sets from existing testing platform are employed to train this model and verify through simulation with MATLAB environment. The results show that the proposed model agrees well dynamically with the LDAC systems. The average testing errors are less than 5% for humidity ratio prediction and less than 10% for temperature prediction, respectively. The presented dynamic model is valuable to further study on dynamic control strategy development of the dehumidifier.