|
|
|
| Interpretable prediction of heat transfer coefficients for direct contact evaporator based on DA-SVM-SHAP hybridization strategy |
|
|
|
|
Abstract The objective is to investigate the influence of various factors on the heat transfer coefficient in a low-temperature metallurgical waste heat recovery system employing a direct contact evaporator. The complexity of the nonlinear relationship makes it challenging to describe the correlation using traditional formulas. This work proposed the use of the dragonfly optimization algorithm (DA) in connection with the support vector machine (SVM) to optimize and achieve intelligent prediction of the direct contact evaporator heat transfer coefficient. The DA-SVM algorithm was capable of powerful fitting ability, which realized the intelligent prediction of the heat transfer coefficient of the direct-contact evaporator with a high degree of accuracy. The findings demonstrate that the DA-SVM algorithm can accept the inlet temperature of heat-conducting oil, outlet temperature of heat-conducting oil, outlet temperature of organic substances, steam flow rate, and heat absorbed by the organic medium as input variables, and the heat exchange coefficient of the direct-contact evaporator as output variables. In comparison to the four prediction models which named the back propagation neural network, extreme learning machine, kernel extreme learning machine, and extreme gradient boosting tree, the root mean square error of the DA-SVM model was reduced by 10.5%, and the coefficient of determination was improved by 2.6%. The extent to which the input variables contribute to the prediction performance of the DA-SVM model is explain by the SHAP value. Hence, the DA-SVM model is capable of accurately predicting the heat transfer coefficient of the direct contact evaporator. Further visualization and analysis demonstrate the degree of correlation between each input variable and the prediction results, which affect the heat transfer coefficient, which illustrates a significant advantage in modeling the nonlinear relationship of the heat transfer process. Based on above, this work offers a robust foundation for further optimization of the operational parameters of the low-temperature waste heat recovery system for metallurgical processes and enhancement of the heat exchange performance of the direct contact evaporator.
|
|
Received: 23 October 2024
Published: 03 April 2025
|
|
|
|
|
|
|
2025, Vol. 46 
