It is well known that reducing variability is the basis of quality control activities. The variability of outcome of the production process, which is composed with product planning, design and manufacturing, can be regarded as the three kinds of variability as follows: 1）the variability before shipping to market, 2）the variability after shipping to market, 3）the variability of satisfaction of market. The actions for reducing the variabilities can be classified and systematized as the four kinds of approach by depending on which elements of the causal relationship an action is taken against. In this article, the systematization of the four kinds of approach for reducing the three kinds of variability is discussed. Moreover, additional discussions on the today’s direction and issues in this systematization under the IoT and AI technologies are shown.

View full abstract

　This paper considers the direction of SQC based on the author’s experience. First, it is described that the environment surrounding quality is changed while introducing the activity report of the Industry-Academia Collaborative Study Group. Next, it is argued that quality management specialists need hot keywords in order to be active in the company in such an environment change, and that the keyword data science is compatible. Finally, this paper proposes the establishment of SQC that incorporates machine learning using data science as a keyword.

View full abstract

　Multi-axis machine tools and coordinate measuring machines have been increasing in machining process.
　On the other hand, international standardization of geometrical product specification (GPS) is also being actively discussed.
　The GPS is index determined to evaluate the product conformity, and they are specifications to conform the target function of the product. The measures to achieve the specification should be considered in each manufacturing process based on their particular technical expertise.
　Manufacturing engineers who establish or improve their process need to use the process data based on different perspective than before. The following points would be useful for the process variation of GPS.
　Form deviation: observe the actual shape and specify the deviation factor.
Orientation and location deviation: use the resolved coordinate values instead of the GPS value to find the cause of the variation.
　In addition, GPS have their hierarchical structure, correlation with other GPS items should be detected.

View full abstract

　This report introduces the 6th research meeting for knowledge sharing of “ Service Excellence Division ” and “ Production Innovation Division ”.
　We focus on and discuss “ transform of business structure with emerging changes reflected ”.

View full abstract

　This proposal describes the checkpoint for the prevention of injustice based on practical quality management system『practical QMS』.
　Organizational intellectual assets and practical quality management system are essential for prevention of injustice by my experience of corporate quality officer. Organizational intellectual assets base on ISO 9001, and ISO 9004 are not sufficient for the prevention of injustice. Process Management based on Japanese Quality Control, TQC, and TQM are essential for the prevention of injustice. And also, ethics and compliance are essential for prevention of injustice.
　There are more root causes of injustices; week management of Quality, poor continual QMS learning, insufficient risk management and no chief quality officer (CQO).
　Next proposal describes checklist for the sustained prevention of justice based on practical QMS.

View full abstract

　Many quality problems have occurred in the Japanese manufacturing industry.
　Brand power of Japanese quality has thus been declining.
　Therefore, the Japanese manufacturing industry is trying to strengthen the organization’s ability of defect prevention and quality control.
　This article introduces how to use the latest IoT technologies in order to enhance the organization’s ability of the defect prevention and quality control.

View full abstract

　When cause-effect relationships can be described as a linear structural equation model，Brito and Pearl (2002) proposed the conditional instrumental variable (CIV) method to identify the total effect．In this paper，when there are several CIVs，we propose the weighted CIV estimators for the total effect with the better estimation accuracy.

View full abstract

　As trials to clarify structure of consumers’ evaluation of product quality, Akiba et al. (1981) obtained eight expectation factors as the viewpoints and Kano et al. (1984) proposed two dimensional recognition model which categorize quality elements of a product into attractive, must-be, one-dimensional quality ones and so on. On the other hand, Miyakawa (1990) utilized x² distance for two-way contingency table which consists of quality elements and attractive, must-be, one-dimensional quality ones and so on in order to explore association structure. In this paper, firstly, correspondence analysis is used for the two-way contingency table for smart phones, and secondly by using analysis for three-way contingency table some new knowledge about the association structure is obtained.

View full abstract

　In the early stage of experiments, there are only little information. So, experimenters have no thoughts about factors’ number, factors’ effects, and a size of error variance. Then experimenters design experiments using orthogonal array. It needs preferably an error variance with more than about five or six degrees of freedom, but it is often less than this, because experimenters would like to minimize the number of experiments.
　When performing analysis of variance about observed data by using three-level orthogonal array, called Chi-square plot with two degrees of freedom was proposed. There are practically no big problems from the results of power of test using numerical simulations.
　This revised method leads to promote and support applications of three-level orthogonal array to practical engineering affairs.
　The advantages to usual analysis of variance are as follows.
1. This method of pooling free can be easily used and needs no further understandings about mathematical statistics and design of experiments.
2. This method has a possibility for finding some significant factors which aren’t being considered.
3. This method can be used even when error variance is unknown, and it’s possible to estimate error variance.
4. The power of test of this method is almost equal to usual analysis of variance by F test.

View full abstract

　Most organizations address to the activities of preventing recurrence of troubles or accidents. However, they often fail to identify the weakness in use of organizational knowledge on causes and countermeasures behind them and to prevent the troubles or accidents caused by similar causes. This study focuses on troubles or accidents due to human inappropriate behaviors in production or service provision, and proposed a model that presents depth and type of cause and its countermeasure from the viewpoint of Root Cause Analysis, as well as a method for evaluating the recurrence prevention activity using four indexes that are obtained by comparing recurrence prevention reports with the model. It also applied this method to 20 actual equipment-used production processes. As the results, about 70% of the processes judged that they could complete the evaluation within one day and there would be no variation if experienced persons performed evaluation. It was also found that the deeper the direct and root causes are analyzed and the countermeasures are taken (the higher the evaluation values I and II), and if deeper countermeasures are taken, the more the depth of analysis and the depth of countermeasure as well as the type of cause and the type of countermeasure are matched (the lower the evaluation values III and IV), the lower the rate of trouble/ accident recurrence.

View full abstract