Quality Engineering deals with all kinds of noises. Earthquakes are big nature noises. Designing robust houses than competaters is a role of quality engineering. However what kind of noise we should pick up is the most important step from the viewpoint of robustness test. In this article it is discussed not only big nature noise but also explosion by a criminal person. In all business fields there is monetary transaction and there are always monetary problems due to criminal activities.

The designers of public telephone sysytem in ECL suffered from the users who did not pay the fare but also change system. They had to change design everytime by the system after knowing the noises. The noises by criminal people in information systems would be the most important noises including in 21st century. From the viewpoint of quality engineering the measure of functionaly needs not changes, but we have to predict the noises by climinal people.

Today, eggs are considered a good food because of the low price and availability in large quantity. Our company is producing and selling thick baked eggs under the name "ltamagoyaki". The product is sold to stores or customers to be put together with sushi or bento(such as a lunch box).

It is made through a production line that includes the processes of mixing,baking, packaging and sterilizing. It is said that 70 to 80 percent of taste (savoriness) of this sponge-like product is contributed by physical properties rather than chemical properties. In this study, technology development was conducted to improve the physical property related to mechanical measurement instead of using sensory test.

In a milling process, it is difficult to optimize precision, tool life and processing efficiency together in one study. It was necessary befbre this study to adjust the process conditions every time based on the objective of processing a certain product. In this study, the two-stage design method was used by conducting two separate experiments. In the first stage, transformability was used for optimization by conducting an off-line experiment. In the second stage, a simple life test was conducted. The tool life and processing efficiency were converted monetarily using the Loss Function. After the optimization, it became possible to select the best cutting speed based on the objective of processing, and determine the relationship between tool life and processing efficiency.

Automatic transmission is a device to automatically transfer the power from the engine of a car to the driving wheels with different speed ratios. It is desirable that the transferring be comfortable, smooth, and low in fuel consumption.

Inside the automatic transmission, there are many clutch mechanisms assembled as the elements for speed changes. These clutches are attached with a wet friction material required to have a good friction characteristic and stability. In this study, parameter design was applied to develop the friction material early in the development, ie., before the development of new types of automatic transmissions. From the study, the friction material with a good functionality was developed, together with many successful results.

損失関数(QLF:quality loss function)を適用した許容値設定はロバスト設計における重要な理念の一つである。この論文では,MTS/MTGSを用いた多次元情報システムのためのしきい値設定における損失関数の役割について論じる。しきい値は多次元情報を用いた診断を有効に行うために非常に重要である。本論文では,いくつかの異常状態に対する損失関数によるアプローチを例と共に明示する。従来の多変量解析による方法では,しきい値は誤判別により予想されるコストと確率を用いて決定されてきた。MTS/MTGSでは異常状態は独立した状態として扱われないので,こうした処理は適用されない。

The ability of a programmer is an important factor that directly affects the quality of software. In this study, a method to forecast the programmer's ability from the questionnaire asking the programmer's life style using MTS approach. 153 examinees participated to write simple programs and answered the questions regarding their life style. From the results, a base space and the Mahalanobis distance from the abnormal data was calculated using Schmidt's orthogonal expansion.

As a result, it was concluded that regarding the judgment of the direction of abnormality(i.e., abnormal in good direction or abnormal in bad direction), it is better to use individual item rather than using the whole items in the questionnaire. In other words, SN ratio was used to evaluate the linearity of individual item and select those items with high linearity. But doing this way, it became possible to forecast the programmer's ability.

A system for classifying images in the presence of noise was developed and optimized using Robust Design methods. The image recognition system uses wavelet transforms to extract features from images. These features were used to construct Mahalanobis spaces for each type of image in a set. The system classifies noisy images by comparing the Mahalanobis distance to all of the image types in the set and selecting the image type with the smallest distance. The system was tested using gray-scale bitmaps of four famous portraits. Robust Design methods were employed to optimize the selection of image features used to construct the Mahalanobis space. The optimized system employs only 14 coefficients for classification and correctly classifies more than 99% of the noisy images presented to it.