Abrupt change of process behavior: The Anderson-Darling detection tool

ABSTRACT

A manufacturing process must be stable in order to produce a quality product. When a process-stabilizing factor fails, the behavior of process data, which should indicate a change has occurred, also changes. This change may be discovered by different methods, among which the Anderson-Darling (AD) method has certain advantages, when the change relates to the nature/shape of the data distribution, while location and scale parameters remain unchanged. The authors propose a method of the change point detection, based on tracking the maximal contrast between AD statistics before and after the sliding time point inside a moving data window. For the purpose of easy and effective use of the proposed method, a free access auxilary Excel™ program tool was developed. Performance of the proposed method is studied and two illustrative examples of its application are presented. The proposed tool can be used both for online and offline data processing.

KEYWORDS:

• change point estimation
• fault detection
• goodness-of-fit methods
• process monitoring
• simulation modeling

Acknowledgments

The authors express their deep gratitude to Prof. Shuki Dror and Dr. Sergey Mazin for their continued interest in the work.

We would like to express our deep gratitude to the anonymous reviewers for the mission they undertook to attentively read this article and their remarkable professional comments that allowed, as we hope, to significantly strengthen and improve our article.

Notes

1 http://www.braude.ac.il/faculty/YarivMarmor.aspx

Additional information

Notes on contributors

Y. N. Marmor

Y. N. Marmor, Ph.D., is senior lecturer of the Industrial Engineering & Management Department at the ORT Braude College of Engineering. His research focus is in healthcare industry. He is the author of peer-reviewed papers in Industrial Engineering and in Healthcare journals and conference proceedings. His current interests involve data-based decision making in different fields of healthcare quality improvement.

E. Bashkansky

E. Bashkansky, D.Sc., is professor emeritus of the Industrial Engineering & Management Department at the ORT Braude College of Engineering. The focus of his present teaching and research activity is quality engineering and metrology. He is the author of 80 peer-reviewed papers in physics and engineering, which have been published in different scientific journals and conference proceedings. He is a member of ISO/REMCO subcommittee, international projects, and conference program committees, founder and a permanent chair of the annual Galilee quality conference. In 2014, he received the Certificate of Appreciation from Israeli Society for Quality for his contribution to the advancement of quality in Israel.