ZIF-8材料的吸附结合人工神经网络的可视化分析  

Adsorption performance of ZIF-8 material combined with artificial neural network visualization analysis

作　　者：谢斐然 季厚坤 洪星悦 鲁亮 张晓娟 XIE Fei-ran;JI Hou-kun;HONG Xing-yue;LU Liang;ZHANG Xiao-juan(School of Chemistry and Chemical Engineering,Key Laboratory of Agricultural Resources and Environment in High Education Institute of Guizhou Province,Anshun University,Anshun 561000,China)

机构地区：[1]安顺学院化学化工学院/贵州省高等学校农业资源与环境研究重点实验室,贵州安顺561000

出　　处：《化学工程师》2025年第1期19-22,79,共5页Chemical Engineer

基　　金：贵州省高等学校农业资源与环境研究重点实验室项目(黔教技[2023]025号、安市科平[2023]5号);安顺学院大学生创新创业训练计划项目(2024106670026)。

摘　　要：以Zn(NO_(3))_(2)·6H_(2)O为原料,二甲基咪唑为配体,通过室温合成法制备了沸石咪唑酯骨架ZIF-8,利用FTIR、XRD及SEM对ZIF-8的结构、形貌进行表征分析,并以刚果红为吸附对象研究其吸附性能。结合现有实验数据构建全连接神经网络预测系统,利用Re LU函数激活引入ZIF-8不同投加量、溶液浓度和pH值与刚果红去除率之间的非线性关系,Adam优化器和MSE损失函数可有效训练模型及优化性能,从而实现ZIF-8对刚果红的吸附预测可视化分析。The zeolite imidazole ester skeleton ZIF-8 was prepared by room temperature synthesis using zinc nitrate hexahydrate as raw material and dimethylimidazole as ligand,and the structure and morphology of ZIF-8 were characterized and analyzed by FTIR,XRD and SEM,and its adsorption performance was studied with Congo red as the adsorption object.Combined with the existing experimental data,a fully connected neural network prediction system was constructed,and the nonlinear relationship between different dosages,solution concentrations and pH value of ZIF-8 and the removal rate of Congo red was introduced by the ReLU function,and the Adam optimizer and MSE loss function could effectively train the model and optimize the performance,so as to realize the visual analysis of ZIF-8 adsorption prediction of Congo red.

关 键 词：ZIF-8 吸附 刚果红 全连接神经网络 染料去除 

分 类 号：TS193[轻工技术与工程—纺织化学与染整工程]