人工智能技术在防腐涂料研发中的应用研究  

Research on Application of Artificial Intelligence Technology in Development of Anticorrosive Coatings

作　　者：王亚鑫 曹亚成 狄志刚 马智俊 史立平 谭伟民 杨小渝[3,4] 马菱薇 张达威 WANG Yaxin;CAO Yacheng;DI Zhigang;MA Zhijun;SHI Liping;TAN Weimin;YANG Xiaoyu;MA Lingwei;ZHANG Dawei(CNOOC Changzhou Paint&Coatings Industry Research Institute Co.,Ltd.,Changzhou,Jiangsu 213016,China;National Engineering Research Center for Coatings,Changzhou,Jiangsu 213016,China;Computer Network Information Center,Chinese Academy of Sciences,Beijing 100083,China;University of Chinese Academy of Sciences,Beijing 100049,China;Institute of Advanced Materials&Technology,University of Science and Technology Beijing,Beijing 100083,China)

机构地区：[1]中海油常州涂料化工研究院有限公司,江苏常州213016 [2]国家涂料工程技术研究中心,江苏常州213016 [3]中国科学院计算机网络信息中心,北京100083 [4]中国科学院大学,北京100049 [5]北京科技大学新材料技术研究院,北京100083

出　　处：《涂料工业》2025年第3期1-6,12,共7页Paint & Coatings Industry

基　　金：国家自然科学基金(62376258,U24B20126);中国海洋石油集团有限公司项目(KJZH-2025-0025)

摘　　要：针对现有防腐涂料研发成本高、研发效率低的问题,通过数据输入、特征选择、测试集比例选择、机器学习算法种类选择、模型训练/评估等步骤,训练出可进行耐老化性能预测的人工智能(AI)模型。研究了不同模型参数对AI模型预测效果的影响,得到了适宜的模型参数。在AI模型复用及验证的过程中,创新性地采用“AI预测+正交设计”的方法进行新型防腐配方优化设计和验证,相较基于经验试错法、正交试验法,采用新方法所需的实验量分别减少93%、78%,显著提升新型涂料研发效率,降低研发成本,有助于解决材料智能研发面临的“小样本”难题。In response to the problems of high research and development costs and low research and development efficiency of existing anti-corrosion coatings,an artificial intelligence(AI)model capable of predicting aging resistance performance was trained through steps such as data input,feature selection,test set ratio selection,machine learning algorithm type selection,and model training/evaluation.We studied the influence of different model parameters on the prediction performance of AI models and obtained suitable model parameters.In the process of reusing and validating AI models,an innovative approach of“artificial intelligence prediction+orthogonal design”was adopted for the optimization design and validation of new anti-corrosion formulas.Compared with the research methods based on empirical trial and error and orthogonal experiment,the experimental requirements of the new method were reduced by 93%and 78%respectively,significantly improving the efficiency of new coating research and development,reducing research and development costs,and helping to solve the"small sample"problem faced by intelligent material research and development.

关 键 词：防腐涂料 研发效率 机器学习 人工智能模型 耐老化性 

分 类 号：TQ630.7[化学工程—精细化工]