Method for predicting salt rejection and energy consumption of flow capacitive deionization technology based on machine learning

The prediction model constructed using XGBoost and SHAP methods solves the problems of low accuracy and lack of interpretability in the desalination rate and energy consumption of FCDI systems, achieving high-precision prediction and parameter optimization, and promoting the engineering application of seawater desalination technology.

CN122245519APending Publication Date: 2026-06-19SOUTH CHINA NORMAL UNIV
0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
SOUTH CHINA NORMAL UNIV
Filing Date
2026-03-11
Publication Date
2026-06-19

AI Technical Summary

⚠Technical Problem

In existing technologies, FCDI systems have low prediction accuracy for desalination rate and energy consumption, weak model generalization ability, and lack of feature interpretation ability, resulting in reliance on physical experiments and lack of basis for parameter optimization, making it difficult to achieve efficient seawater desalination.

⚗Method used

By employing the Extreme Gradient Boosting (XGBoost) model combined with the Shapley Additive Interpretation (SHAP) method, and through data preprocessing, model training and validation, the influence of features is quantified, parameters are optimized, and a high-precision, interpretable prediction process is constructed.

🎯Benefits of technology

This study achieved high-precision prediction of desalination rate and energy consumption in FCDI systems, reduced experimental costs, improved model generalization ability, provided a scientific basis for parameter optimization, and promoted the engineering application of seawater desalination technology.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure 1
    Figure 1
  • Figure 2
    Figure 2
Patent Text Reader

Abstract

This invention discloses a method for predicting desalination rate and energy consumption of FCDI (Fluorescent Capacitive Deionization) systems based on extreme gradient boosting and Shapley additive interpretation. The method includes: collecting 10-dimensional characteristic experimental sample data of a FCDI system; after normalization and Pearson correlation coefficient matrix analysis to remove redundant features, dividing the training and test sets in an 8:2 ratio; constructing an extreme gradient boosting prediction model for desalination rate and energy consumption; determining the optimal hyperparameters and completing training through a grid search method combined with five-fold cross-validation; and verifying the model performance using the coefficient of determination (R²), mean absolute error (MAE), and root mean square error (RMSE). This invention offers high prediction accuracy and strong generalization ability, solving the "black box problem" of traditional machine learning models, significantly reducing the experimental cost of FCDI technology research, providing a scientific basis for FCDI system parameter optimization, and can be widely applied to the design, operation, and optimization of FCDI systems in seawater desalination engineering. It can also be extended to traditional capacitive deionization, electrodialysis, and other seawater desalination technologies.
Need to check novelty before this filing date? Find Prior Art