化工学报2026，Vol.77Issue(5)：2307-2321,15.DOI:10.11949/0438-1157.20251297
机器学习在结晶过程建模与控制中的研究进展
Recent advances in machine learning for modeling and control of crystallization processes
摘要
Abstract
Crystallization is a fundamental solid-liquid separation operation in chemical,pharmaceutical,food and materials industries.Achieving high product quality requires precise control of critical process parameters such as temperature,feed rate and solvent composition.Conventional modeling and control strategies that rely on mechanistic formulations or empirical correlations often fail to represent the highly nonlinear and multiscale nature of crystallization systems.In recent years,machine learning,with its powerful data-driven feature extraction and pattern recognition capabilities,has provided new ideas for the intelligent research of crystallization processes.This review summarizes recent developments in machine learning enabled process monitoring,predictive modeling and control optimization for crystallization.The potential of these methods to enhance process characterization,improve model fidelity and support intelligent decision making is highlighted.Challenges related to data quality,model generalization and reliable industrial deployment are also discussed.关键词
结晶/神经网络/过程分析技术/监测技术/模型预测控制/粒度分布Key words
crystallization/neural networks/process analytical technology/monitoring techniques/model-predictive control/size distribution分类
信息技术与安全科学引用本文复制引用
闫艺航,马渊,刘程琳,于建国..机器学习在结晶过程建模与控制中的研究进展[J].化工学报,2026,77(5):2307-2321,15.基金项目
上海市自然科学基金项目(25ZR1401083) （25ZR1401083）
江西省重点研发计划项目(20223BBG74008) （20223BBG74008）
