Evaluation of interval regression analysis for uncertain resistance spot welding quality data

ABSTRACT

This article presents an interval regression analysis method to model uncertain resistance spot welding (RSW) quality data. The goal of this approach is to support making decisions using the RSW quality data (e.g. determining welding conditions on a combination of materials). The RSW data include various welding conditions where each welding condition represents material characteristics and welding process parameters associated with uncertain welding quality. Our experiments reveal that interval regression analysis is capable of mapping welding conditions to their uncertain welding quality. Due to its high uncertainty in the RSW quality data, the interval regression models are unable to provide high fitness measure. However, these models provide advantages, including the capability of providing valuable insights to analyse the uncertainty and support making decisions.

KEYWORDS:

• Interval regression analysis
• resistance spot welding
• manufacturing process data
• data-driven decision support
• data uncertainty

Acknowledgements

This paper is extended from the authors’ FAIM 2016 conference paper, titled ‘Making Data-driven Decision for Uncertain Resistance Spot Welding Process Data Using Interval Regression Analysis.’ This research is partially supported by the US National Science Foundation Industry & University Cooperative Research Center for e-Design (IIP-1338780) and Ford Motor Company.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This research is partially supported by the USA National Science Foundation Industry & University Cooperative Research Center for e-Design (IIP-1338780) and Ford Motor Company