Abstract In this study, different machine learning (ML) models has been investigated for DCMD process and provide an accurate and quick modeling tool. Direct contact membrane distillation (DCMD) modules may perform better if fast and accurate modeling techniques are developed to forecast their performance. Three machine learning models (MLR, KNN, and XGBoost) have been thoroughly tested to assess their resilience in computing time, trend predictability, and model accuracy. Findings demonstrate that the XGBoost and MLR models outperform the KNN model in terms of performance, with corresponding MAPE test values of 2.07 %, 3.36 %, and 3.53 %, respectively. Additionally, ML models have the advantages of increased accuracy, reduced computational time, and simplicity compared to conventional models. The feed temperature is one of the operating parameter that affects DCMD permeate flux the most, according to the feature importance analysis.
Kumar et al. (Thu,) studied this question.