タグチメソッドの精緻化の研究

田口方法的详细研究

• 批准号: 22K11942
• 负责人: 永田 靖
• 金额: $ 2.33万
• 依托单位: Waseda University
• 依托单位国家: 日本
• 项目类别: Grant-in-Aid for Scientific Research (C)
• 财政年份: 2022
• 资助国家: 日本
• 起止时间: 2022-04-01 至 2025-03-31
• 项目状态: 未结题
• 来源: https://kaken.nii.ac.jp/grant/KAKENHI-PROJECT-22K11942/
• 关键词: タグチメソッド; MTシステム; RT法; 混合モデル; アンケート調査データ; 異常検知; 予測; ロバスト推定; 機械学習

MTシステムに関連し，次の研究を行った．①正常空間，既知の異常がある場合に，判定すべきデータが正常であるか，既知の異常であるか，未知の異常であるかを判定する手法を開発し，性能評価した．この手法は混合モデルに基づくものである．さらに，教師あり学習と半教師あり学習の両者を考慮し，性能の比較を行った．成果を国際会議（Asian Network for Quality）で発表した．この発表はBest Paper Awardに選出された（2021年度）．2022年度では，その内容に，さらに改良を加えて学術誌に投稿し，掲載された．②アンケートでは，５件法や７件法でデータを採取する．したがって，背後には連続的な計量空間が想定されるが，データ自体は離散量が得られる．このような場合，離散量に基づく相関関係は希薄化されてしまう．したがって，そのような状況で異常値を検出するためには，工夫が必要になる．そこで，連続変数と順序尺度変数間の相関として多分系列相関係数をMT法に組み入れて異常検知する方法を開発した．成果を国際会議（Asian Network for Quality）で発表した(2021年度)．2022年度では，その内容に，さらに改良を加えて学術誌に投稿し，掲載された．③高次元データに対応できるRT法を定式化して性能を比較した．代表的な高次元主成分分析法であるノイズ掃き出し法とクロスデータ行列法を取り上げ，それらを主成分分析を用いたタグチのRT法に導入し，その異常検出性能について検証した．成果を国際会議（Asian Network for Quality）で発表した（2021年度）．2022年度では，その結果に検討を加えて学術誌に投稿し，掲載された．

In MT, we need to study the normal space. 1. In the normal space, we know that we know that we are in good condition, and that we know that we know that we are in good condition, and that we know that we know that we have a lot of trouble, and that we know that we know that we have a lot of trouble, and that we know that we know that we have a lot of trouble, and that we know that we know that we have a lot of trouble. The results of the International Conference of the International Conference (Asian Network for Quality). The results of the International Conference on International Conference (Asian Network for Quality). The results of the International Conference (International Conference). The results of the International Conference (International Conference). The results of the International Conference (International Conference). The results of the International Conference (International Conference). The results of the International Conference (International Conference). The results of the International Conference (International Conference). The results of the International Conference of the International Conference (Best Paper Award). There are 5 parts, 7 pieces of this method, and 7 pieces of this method. In this way, the calculation space of the back-to-back contact volume is determined, and the self-emission volume of the device is in good agreement with each other. the bulk of the gas is used to reduce the cost of the product, and it is necessary to use it in a short time. Linked data sequence scale series correlation multi-series correlation MT method organization to learn about the method to start the competition. Results International Conference (Asian Network for Quality) presentation Table (2021). 2022 Annual meeting, contents of the Conference, Library improvement and Log Log contribution Program The higher dimensional principal component analysis (PCA) method, which is represented by the higher dimensional principal component analysis (PCA) method, represents the higher dimensional principal component analysis (PCA) method, which is represented by the higher dimensional principal component analysis (PCA) method, which is represented by the higher dimensional principal component analysis (PCA) method. The higher dimensional principal component analysis (PCA) method is based on the determinant method, and the principal component analysis (PCA) is imported by the RT method The results of the International Conference of the International Conference (Asian Network for Quality) will be reviewed in 2021. The results of the 2022 meeting will increase the number of contributions and contributions to the journal.

项目成果

A Consideration of the Recognition Taguchi Method Using High-Dimensional Principal Component Analysis

基于高维主成分分析的田口识别方法的思考

• 发表时间: 2021
• 作者: Ryo Asano;Masato Ohkubo and Yasushi Nagata
• 通讯作者: Masato Ohkubo and Yasushi Nagata

Adaptation of the Mahalanobis–Taguchi Method to Ordinal Scale

马哈拉诺比斯-田口方法对序数尺度的适应

• DOI: 10.17929/tqs.8.14
• 发表时间: 2022
• 期刊: Total Quality Science
• 作者: IWAMOTO Hiroki;NAGATA Yasushi
• 通讯作者: NAGATA Yasushi

Mahalanobis-Taguchi Method for Anomaly Detection and Classification

用于异常检测和分类的 Mahalanobis-Taguchi 方法

• 发表时间: 2021
• 作者: Kentaro Honma;Masato Ohkubo and Yasushi Nagata
• 通讯作者: Masato Ohkubo and Yasushi Nagata