Prediction of a solid desiccant dehydrator performance using least squares support vector machines algorithm

Abstract

This study presents the potential of least squares support vector machines (LSSVM) modeling approaches to predict the moisture content of natural gas dried by calcium chloride dehydrator units. Genetic algorithm (GA) as population based stochastic search algorithms were applied to obtain the optimal LSSVM models parameters. The results revealed that the GA-LSSVM are capable of capturing the complex nonlinear relationship between the input and output variables. For the purpose of predicting water content of natural gas for freshly recharged conditions, the GA-LSSVM model yielded the mean absolute error (MAE) and coefficient of determination (R2) values of 2.7898 and 0.9986; for the whole data set, while for the purpose of predicting water content of natural gas prior to recharging conditions, the GA-LSSVM models yielded the MAE and R2 values of 1.1044 and 0.9995; for the whole data set. Proposed model provides fairly promising approach for predicting the approximate moisture content of natural gas dried by calcium chloride dehydrator units for both freshly recharged and just prior to recharging conditions.