机器学习在超临界二氧化碳制膜技术教学中的应用实践

Application of machine learning to the teaching of membrane technology with the assistance of supercritical carbon dioxide

[目的]探索将超临界二氧化碳(CO_(2))辅助制备金属-有机框架(MOF)膜与机器学习技术相结合的创新教学模式,以提高本科生在实验数据处理、分析和预测方面的能力。[方法]在教学过程中,学生首先使用数据提取工具从复杂的实验图表中提取CO_(2)在MOF材料上的吸附等温线数据,构建包含数据点和材料性能参数的数据集。随后,学生应用极致梯度提升(XGB)和随机森林(RF)两种机器学习算法对提取的数据集进行训练和测试,建立材料性能预测模型。[结果]数据提取工具显著提高了数据处理的效率和准确性,XGB模型在材料性能参数预测方面表现出良好的准确性和可解释性。通过这种结合数据提取和机器学习的教学模式,学生在数据处理和预测方面取得了显著进步,增强了他们的数据分析和科研能力。[结论]本工作提出的教学模式为化学工程和材料科学领域的实验数据处理和分析提供了有效的解决方案,具有较高的应用价值。

[Introduction]An innovative teaching mode combining the supercritical carbon dioxide(CO_(2))assisted preparation of metal-organic framework(MOF)membranes and machine learning technology was explored to enhance the undergraduates’abilities in experimental data processing,analysis,and prediction.[Method]During the teaching process,students use a data extraction tool firstly to obtain CO_(2) adsorption isotherms from complex experimental graphs,creating a dataset containing data points and material properties.Subsequently,students apply extreme gradient boosting(XGB)and random forest(RF)algorithms to train and test the dataset,establishing the predictive models for the properties of materials.[Result]The data extraction tool significantly improves the efficiency and accuracy of data processing.The XGB algorithm shows good accuracy and interpretability in predicting the material properties.Through this teaching mode,which combines data extraction and machine learning,students achieve significant progress in data processing and prediction,enhancing their data analysis and research capabilities.[Conclusion]The teaching mode proposed in this paper provides an effective solution for experimental data processing and analysis in the fields of chemical engineering and material science with high application value.