In recent years, quality demands have become increasingly stringent. Demands on the order of ppm and even demands for no defects at all are taken for granted,but means for assuring ppm-order quality have yet to be established. Prior attempts to establish such means have focused on process management and inspection, but when management and inspection are combined they become more intricate and complex, and the independence of quality assurance from process management is lost. These problems then become a major obstacle to productivity improvement. They seem to be rocking the foundation of the Japanese management, which is based on the simultaneous pursuit of quality and improvement (cost reduction). Working from the idea of producing products of uniform quality, this study attempts to establish a ppm-order quality assurance system by inspecting each product to see that it is identical to a known good product, utilizing the MT method of quality engineering. It presents new ideas for ppm-order quality management through ensuring the independence of quality management and quality improvement activities from quality assurance and applying ideas of on-line quality engineering and parameter design.
Cutting processes always generate chips, which must be cleared away lest they find their way as foreign objects into finished products. In this study, the parameter design techniques of quality engineering were applied to optimize the shape of a cutting tool so as to reduce chips to tiny size, and optimize the washing process performed to clear the chips. Experiments were carried out efficiently, and effectiveness was demonstrated.
In the Mahalanobis-Taguchi (MT) system known as the new pattern recognition methodology, Taguchi method(1), in which the results of the synthetic judgments are able to have plus values or minus values, do not generate multi-collinearity. In this study, we propose to measure the educational effectiveness by Taguchi method(1) in consideration of the characteristics (simultaneity,heterogeneity) of education, and its usefulness is examined. This paper shows the concept of modeling the measurement of educational effectiveness, and proposes some measures for the increase of internal knowledge and compares them. The extension of Taguchi method(1) to the measurement of educational effectiveness is studied, the selection of the characteristic items is examined and the combined estimates are obtained. It is shown that Taguchi method(1) is valid for the measurement of educational effectiveness. As a result, we could extract educational and learning factors for increasing internal knowledge and make improvement proposals from the standpoints of both teacher and student. In addition, we showed that we could make improvement proposals in consideration of the heterogeneity for the individual student.
In a study of a new two-dimensional measurement scale configured with a commercial scanner and an in-house analysis program, a functionality evaluation of combined standard uncertainty and economic loss due to measurement was performed per JIS Z 9090-1991 : Measurement - General rules for calibration system. As a result, the effect of the new scale was quantified as a loss reduction of 36 yen per unit as compared to the current system. On the basis of this loss reduction, an evaluation of plant and equipment investment in this measurement scale was carried out in a prototype pass/fail test system used to inspect 400 units per year manufactured for in-house delivery, and the net present value method was used to make a decision on plant and equipment investment. From the prototype test system it was found that investment in the new system would exceed expected future gain, and the inappropriateness of the plant and equipment investment was clearly quantified in economic terms. In a case study applied to volume production inspection with a high production volume,the expected future payback exceeded the cost of investment in the new system,and it was shown that again exceeding the investment could be expected.
Among the computational methods used in the Mahalanobis-Taguchi system,the MT and RT methods are used for pattern recognition in classification tasks. Genichi Taguchi proposed the standardized variation pressure method five years before proposing the RT method, but while the standardized variation pressure method went largely unnoticed, when the RT method was presented it attracted much attention. To overcome the problem that the RT method cannot deal directly with data of different dimensions, data standardization was studied,leading to results equivalent to the variation pressure method. In this report, distances given by variation pressure and medical diagnoses derived from these distances will be described in comparison with the MT method. Diagnostic methods using variation pressure will also be mentioned.