The downsizing of a copier necessitated the mechanical redesign of its sidemounted multiple sheet inserter. In this study, control factors were assigned to an L18 orthogonal array, various CAE models of the multiple sheet inserter were constructed, and the models were evaluated by CAE mechanical analysis and parameter design. Specifically, experiments were carried out with the inserter shutting speed and other customer conditions as noise factors and the value of the torque integrated over the operating angle as a characteristic value. In the analysis of the signal-to-noise ratio, the standard S/N ratio was used to tune the design toward the optimal state in which the inserter would operate as smoothly as possible under robust conditions with high noise immunity,and a set of design conditions for a nearly ideal characteristic curve was obtained. The time and cost of the evaluation were reduced by 60% in comparison with experiments on actual machines. The design of the copier was pursued on the basis of the information gained through this study, no problems arose during factory tests, and the copier was placed on the market as a commercial product.
Product planning precedes product development and has a major effect on the success of the developed product, but plans are often not properly assessed, and conventional assessments fail to provide information for reviewing the plan. This information is therefore obtained by techniques such as two-factor cross analysis, but as the number of factors increases this takes considerable time and effort, and if the number of factors is reduced or if attention is paid only to mean effects, the problem of oversimplified targets arises. The present study used a method of assessment by pattern recognition, under the assumption that each person is different. This enables all factors to be assessed at once, with consideration given to relations with other factors, and provides a simple way to assess the all-round fitness of the plan. Since all feature information is used,diversity is present and it is possible to segment customers with complex feature patterns. The objective of this study was to create a plan assessment methodology that would be able to assess the effects of a large number of feature factors simultaneously by pattern recognition. It was found that this enabled the immediate acquisition of feature infbrmation for anticipated customer strata not fitting the plan, and the efficient acquisition of effective information for reviewing the plan.