Implementation and validation of an artificial neural network for predicting the performance of a liquid desiccant dehumidifier

In the current paper, an artificial neural network (ANN) model for predicting the performance of a liquid desiccant dehumidifier in terms of the water condensation rate and dehumidifier effectiveness is proposed. Six air and desiccant inlet parameters were used as inputs for the ANN. To determine the performance of the ANN technique for predicting the performance of the dehumidifier, actual experimental test data were obtained from a previous study, that tested a packed column dehumidifier with a total height of 0.6 m and a specific packing material surface area of 77 m2/m3, using triethylene glycol as the desiccant. In the experiment, 54 data samples were used in a series of runs. MATLAB code was designed to study feed forward back propagation with traingdm, learngdm, MSE, and tansig as the training, learning, performance, and transfer functions, respectively. Up to 70% of the experimental data was used to train the model; the remaining 30% was used to test the output. The results show that the 6-3-3-1 network structure was the best model for predicting water condensation rate, whereas the 6-6-6-1 network structure was the best model for predicting dehumidifier effectiveness. The maximum percentage difference between the ANN and experimental value for water condensation rate and dehumidifier effectiveness were 8.13% and 9.0485%, respectively. The model for the water condensation rate and dehumidifier effectiveness could be further improved by modifying the number of hidden layers.

► We used ANN model for predicting the performance of the desiccant dehumidifier.
► The 6-3-3-1 network structure was the best model for predicting water condensation rate.
► The 6-6-6-1 network structure was the best model for predicting the effectiveness.
► The maximum percentage difference between the ANN and experimental does not exceed 9.0485%.