Studies on parameter and tolerance designs in off-line quality control

Abstract

The Taguchi idea of parameter and tolerance designs has been received considerable attention from statisticians and quality engineers. However, it is only recently that some of the underlying principles of his idea have been clarified and other alternative procedures have been proposed. This thesis reviews the currently available parameter design procedures, and advantages and disadvantages of each procedure are investigated to provide guidelines for choosing an appropriate procedure for a given parameter design problem. In addition, previous parameter design procedures are extended to include the cases of fairly general transfer and loss functions, and the concept of adjustability. Taguchi``s tolerance design consists of two parts: tolerance determining and grade selection procedures. Taguchi``s tolerance determining procedure is extended to handle the case of nonlinear relationship between the characteristics of a product and low-level parts. This thesis also deals with two problems for selecting an appropriate alternative for each low-level parts from an economic point of view. The objective of the first problem is to minimize the quality loss and the manufacturing or assembly cost. In case where the price of a product is several times the unit manufacturing cost, the customer may suffer a loss which is several times the unit manufacturing or assembly cost. For such a case, we formulate the second problem in which the objective is to minimize the quality loss while the manufacturing or assembly cost is treated as a constraint. Heuristic search procedure is also provided to solve the above two problems.