Research on prediction of coal water medium separation effect based on multi-models

Abstract

To improve the separation efficiency of raw coal and ensure clean use, the accurate calculation of the partition coefficients (PCs) in coal water medium sorting is required. Single models have been used to predict the partition coefficient (PC) for decades, but their accuracy remains constrained. This study proposes a multi-model (MM) calculation method based on the Gompertz model (GM), the Logistic model (LM), the Arctangent model (AM), and the Approximate formula (AFM) to improve the accuracy of the predicted coal water medium sorting PCs. Four groups of coal samples and two specific cases were used to verify the accuracy of the MM calculation method. The PCs of the MM method had a minimal Ef (0.91–8.84), a maximal R2 (0.9648–0.9994), a maximal F-value (199.17–11352.31), and the highest significance of all the models. The MM method was found to be the most suitable of all the models for predicting any coal water medium separation process. Further, when calculating the PC for cleaned coal ash, the separation density of MM is closer to the actual separation density than that of either the GM, LM, AM, or AFM models. The MM method, therefore, produces more accurate results compared to a single model. MM is expected to predict the PC based on the required cleaned coal ash, and then regulate the sorting density to improve the production efficiency.

Keywords

• Water medium separation
• Partition coefficient
• Partition curve
• Multi-model prediction
• Imperfection
• Separation density