Prediction of Coal Ash Flow Temperature Based on Gray Relational Analysis, Support Vector Regression and Genetic...

Area of Science:

• Chemical Engineering
• Materials Science
• Computational Science

Background:

• Coal ash flow temperature is crucial for entrained flow bed gasification efficiency.
• The precise relationship between coal ash composition and flow temperature is not well-established.
• Accurate prediction of flow temperature is essential for optimizing gasification processes.

Purpose of the Study:

• To develop a reliable and accurate predictive model for coal ash flow temperature.
• To investigate the application of machine learning, specifically support vector regression, for this prediction.
• To compare the performance of the developed model against traditional methods like FactSage software.

Main Methods:

• Utilized machine learning models, including various support vector regression (SVR) approaches.
• Developed a hybrid model combining gray relational analysis (GRA) and a genetic algorithm (GA) with SVR.
• Evaluated model performance using metrics such as root-mean-square error (RMSE), mean absolute error (MAE), and average deviation.

Main Results:

• The proposed GRA-GA-SVR model achieved high prediction accuracy for coal ash flow temperature.
• Achieved a root-mean-square error of 28.37, mean absolute error of 19.48K, and average deviation of 1.58%.
• Demonstrated significantly higher accuracy and efficiency than FactSage software, which had an MAE of 93.73K.

Conclusions:

• The developed machine learning model is a viable and effective tool for predicting coal ash flow temperature.
• This approach offers a more accurate and efficient alternative to conventional calculation methods.
• The findings highlight the potential of machine learning in advancing coal chemical engineering applications.