报告题目：1. Artificial Intelligence in Chemical Engineering  

2. A Graph-Based Process Chemometrics Method for Root Cause Diagnosis of Process Disturbances

报告时间：2023年5月25日（周四）14：00

报告地点：C1-437

主讲人：姚远 教授

报告摘要：

1. Artificial Intelligence in Chemical Engineering

This presentation explores the impacts of artificial intelligence (AI) in chemical engineering. The roles of chemical engineers include excellent process manipulation, strategic product development, and rapid industrialization. This presentation discusses the potentials and difficulties of applying AI techniques to these three aspects from the viewpoints of chemical engineers. Furthermore, the critical points of the applications of AI to process control are highlighted and some industrial practices are introduced in brief.

2. A Graph-Based Process Chemometrics Method for Root Cause Diagnosis of Process Disturbances

Nowadays, industrial processes become more and more complicated. The large number correlations and interactions between variables often cause difficulties in identifying root causes of process disturbances. Principal component analysis (PCA) has been a popular multivariate statistical method widely used in process chemometrics to process high-dimensional and significantly correlated data. However, the lack of sparsity and ignorance of process connectivity in PCA modeling often makes the results difficult to interpret. In this work, an edge-group sparse PCA (ESPCA) method was adopted to overcome the shortcomings of the conventional PCA. During model training, ESPCA considers the connections between process variables, which is often revealed by piping and instrumentation diagrams. Meanwhile, it adds sparsity to each loading vector, making each principal component more interpretable. In process monitoring, ESPCA shows a similar disturbance detection performance to the conventional method but significantly outperforms PCA in the isolation step which identifies the groups of process variables critically affected by the detected disturbance. Consequently, ESPCA provides a good basis for root cause diagnosis.

主讲人简介：姚远，现任中国台湾清华大学化工系教授。研究领域包括过程数据分析、过程监控、软测量技术、无损检测技术数据分析以及过程控制等。硕士及本科毕业于浙江大学，控制科学与工程专业。博士毕业于香港科技大学。曾先后获中国台湾清华大学青年教师科研奖、最佳青年研究员称号（中国台湾科技部事务主管部门授予）、I&EC Research 2020最具影响力学者称号。至今，共发表SCI论文100余篇，出版著作两部，授权专利12项，主持（及参与）科研项目60余项。