Ко-кластеризация данных многомерных атрибутов качества для оценки факторов взаимного влияния

Abstract

В статье предлагается метод ко-кластеризации стохастических данных многомерных критических параметров процесса (CPPs) с целью оценки влияния обнаруженных факторов на многомерные атрибуты критического качества (CQAs) продукта на стадии первоначального проектирования процесса производства. Метод представляет новый подход к обеспечению качества продукта, который учитывает проблему ко-кластеризации массивов данных CPPs для определения каузальной связи с CQAs. Используется технология неметрического многомерного шкалирования (NMDS) для определения исходных параметров ко-кластеризации.

Nowadays, competitiveness and efficiency of companies must be continuously improved to face worldwide competitors. Evolutions of computer-intensive technologies development and massive integration of numerical simulations to the design process require new methodologies for numerical design of experiments which use to improve quality of products by taking into account uncertainties in product development. Simultaneous clustering of rows and columns, known as co-clustering, is an important method of two-way analysis of empirical data for practical approaches. The article propose a method of contingency data co-clustering based on multivariate statistical analysis (MSA) for evaluating the influence of critical process parameters (CPPs) factors on the time multivariate critical quality attributes (CQAs) of product. Factorized multivariate CPPs increase the possibility to usemethods of multivariate statistical analysis for evaluating the influence of CPPs on the multivariate CQAs. The article solves the objective of product’s quality assurance at the stage of the initial manufacture process design in accordance with the process-analytical technology for the design of modern certified manufacturing known as "quality-by-design" (QbD). The method proposed in the article presented a new approach of product’s quality assurance which takes into account the block clustering problem on both the individuals and variables parameters for data arrays of computer format. A key feature of the article is the use of nonmetric multidimensional scaling (NMDS) technology to determine the initial parameters of co-clustering. Cluster analysis is an essential tool in different kinds of scientific areas including data mining. Therefore, the subject matter of the article is relevant and given for further development.