Prediction of various parameters of desalination system using BOA- GPR machine learning technique for sustainable development: A case study

Abstract

The present study examines the performance of desalination based atmospheric water extraction system under various climate situations. The Bayesian optimisation for model training hyperparameters was used to make the process autoregressive, and the Gaussian Process Regression (GPR) technique was used to develop the prediction model. Because of its capacity to incorporate data-specific uncertainties and non-linearities, the GPR model was highly efficient in calculating the efficiency of system in different environmental circumstances. For its prognostic efficiency, the model was examined using multiple statistical methods such as R2, mean squared error (MSE), and mean absolute error (MAE). The findings revealed that the GPR model. The prediction model's statistical parameters showed a high prognostic efficiency, with a training R2 of 0.97, MSE of 7282.3, RMSE of 85.34, and MAE of 63.93, and a test R2 of 0.98, MSE of 7596.3, RMSE of 87.16, and MAE of 72.49. Overall, this paper provides a valuable contribution to developing Desalination technology in different climatic regions. The research also sheds light on the relationship between input variables (climatic conditions) and output variables (energy intensity and water extraction rate) for constructing and optimizing desalination based water extraction system for various geographies and climates.