Chance constrained quality control

Abstract

Consideration of quality implicitly introduces the need to adjust inputs in order to obtain desired output. Output evaluation must consider quality as well as cost. Real decisions may involve other objectives as well. Production problems often involve a dynamic situation where the relationship between cost and quality must be experimentally developed. The proposed method is to use regression as a means of identifying input-output relationships, to include variance. A chance constrained multiobjective model can be developed, and decision maker preference incorporated through interactive analysis. Through application of the proposed method, efficient solutions providing as much quality as desired at minimum cost are capable of identification.

Past applications in the area are discussed, as are data collection, modeling, and solution procedures. A number of multiobjective concepts are reviewed in light of the chance constrained model. Constrained techniques are considered more appropriate than weighting techniques for this class of problem. The abilities of currently available solution techniques to support multiobjective analysis for this class of problems are also discussed.