Title of article
An industrial perspective on implementing on-line applications of multivariate statistics
Author/Authors
I. Miletic، نويسنده , , S. Quinn، نويسنده , , M. Dudzic، نويسنده , , V. Vaculik and M. Champagne، نويسنده ,
Pages
16
From page
821
To page
836
Abstract
Multivariate statistics (MVS) has enjoyed popularity in the applied science literature over the last decade. It has also been well
received by industrial practitioners; however, industrial applications have had mixed results. In this paper, we focus on our experiences
in developing multivariate statistical systems for industrial use. From these experiences, we identify a methodology for
developing useful, long-standing industrial applications. We highlight features we feel are important in the successful development
of on-line MVS applications, both technical and non-technical. Specifically, we focus on on-line systems for manufacturing environments.
Many of the applications discussed grew out of industry-university collaborations. To end the paper, we recommend
topics open for further academic research with an industrial focus.
Keywords
Principal component analysis , Partial least squares , Correlation , Fault detection , process operation , Steel manufacturing , Pulp and paper production , Regression
Journal title
Astroparticle Physics
Record number
401431
Link To Document