Abstract
Reverse osmosis (RO) is a key technology for seawater desalination, but boron removal remains challenging due to the relatively low and varying boron rejection of RO membranes. This study explored the use of machine learning (ML) to develop predictive models for boron removal of RO membranes. Data of 11 features encompassing membrane properties, testing conditions and membrane performance were collected from journal articles. Missing data were recovered using data imputation algorithms. The predictive models were developed using five regression algorithms: linear, ridge, decision tree, random forest and XGBoost regressors, and the tree-based XGBoost regressor performed the best (R2 = 0.84). Feature importance analysis and tree diagrams revealed that membrane type, feed pH and NaCl rejection as key factors in influencing boron rejection, while membrane surface properties showed minimal impact. Partial dependence plots were generated to further analyze each feature. High NaCl rejection of >99.6 % is highly desirable for SWRO membranes to achieve high boron rejection. For BWRO membranes at pH >9, a looser structure with a NaCl rejection >95 % could be applied. The study successfully applied ML to a dataset with large portion of missing values, and the results provide valuable insights for future membrane design and boron removal processes.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.