机器学习结合分子指纹预测过硫酸盐氧化体系污染物的降解速率常数  

Machine Learning Combined with Molecular Fingerprinting to Predict the Degradation Rate Constant of Pollutants in Persulfate Oxidation System

作　　者：杨鹰[1] 潘思远 YANG Ying;PAN Siyuan(School of Chemistry and Chemical Engineering,Central South University,Changsha 410083,China)

机构地区：[1]中南大学化学化工学院,湖南长沙410083

出　　处：《徐州工程学院学报（自然科学版）》2022年第2期42-48,共7页Journal of Xuzhou Institute of Technology(Natural Sciences Edition)

基　　金：国家农业科学研究专项基金项目(201503108)。

摘　　要：降解速率是环境污染物处理工艺选用和设计中必须考虑的关键因素,与传统实验方法相比,基于定量构效关系(QSAR)模型预测有机物降解速率的方法可以明显降低人力、物力和时间成本.文章基于分子指纹(MF),分别使用随机森林和极端梯度提升(XGBoost)机器学习算法,成功构建了预测过硫酸盐高级氧化体系中有机污染物降解速率常数的MF-QSAR模型.与MD-QSAR模型相比,MF-QSAR模型在应对更大的数据集时,仍能表现出良好的性能,XGBoost模型的决定系数和均方根误差分别为0.8378和0.3160.最后,使用SHapley Additive exPlanations方法解释了MF具体位点对模型结果的贡献度,XGBoost模型正确区分了富电子和吸电子分子特征对模型的正负影响.Degradation rate is a key factor to be considered in the selection and design of environmental pollutant treatment processes.Compared to traditional experimental methods,predicting the degradation rate of organic matters based on the quantitative structure-activity relationship(QSAR)model can significantly reduce manpower,material consumption,and time costs.In this work,MF-QSAR models for predicting the degradation rate constants of organic pollutants in persulfate advanced oxidation systems were successfully constructed based on molecular fingerprints(MF)using random forest and extreme gradient boosting(XGBoost)machine learning algorithms,respectively.Compared with the molecular descriptors(MD)-based MD-QSAR model,the MF-QSAR model still showed good performance when dealing with larger data sets.The coefficient of determination and root mean square error of the XGBoost model are 0.8378 and 0.3160,respectively.Finally,the SHapley Additive exPlanations method is used to explain the contribution of specific sites of MF to the model.The XGBoost model correctly distinguishes the positive and negative effects of electron-rich and electron-withdrawing molecular features on the model.

关 键 词：降解速率常数 定量构效关系 分子指纹 机器学习 

分 类 号：X703.1[环境科学与工程—环境工程]