Chapter 14: Multivariate Statistical Process Control

Yuan Yao

Department of Chemical Engineering, National Tsing Hua University, Hsinchu, Taiwan

Furong Gao

Department of Chemical and Biomolecular Engineering, Hong Kong University of Science and Technology, Kowloon, Hong Kong, China

In the past several decades, tremendous advancements have been made in automatic process control (APC), which is also called engineering process control (EPC). Although the applications of APC reduce a large part of short-term variations in industrial processes, it is still necessary to detect the abnormal operating conditions and to diagnose the corresponding causes so as to improve the process performance in a long term. Industrial statistics show that even though major catastrophes and disasters are infrequent, minor accidents are very common, resulting in many occupational injuries, and illnesses, and leading to large economic losses every day.1 Therefore, process monitoring and fault diagnosis are essential for ensuring process safety and product quality consistency.

14.1 Statistical Process Control

In process monitoring, a model describing nominal operating conditions based on which an in-control region (normal operating region) can be defined is usually required. Comparing the real status of a process with the model, process abnormalities can be detected as a significant departure from the in-control region. Generally, a process can be modeled based on first principle, expert knowledge, or historical ...

Get Computer Modeling for Injection Molding: Simulation, Optimization, and Control now with the O’Reilly learning platform.

O’Reilly members experience live online training, plus books, videos, and digital content from nearly 200 publishers.