Maintaining model efficiency, avoiding bias and reducing input parameter volume in compressor fault classification

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

With the exponential growth in data collection and storage and the necessity for timely prognostic health monitoring of industrial processes traditional methods of data analysis are becoming redundant. Big data sets and huge volumes of inputs give rise to equally massive computational requirements. In this paper the differences in input parameter selection using a subset of the original variables and using data reduction techniques are compared. Each selection procedure being illustrated by both statistical methods and machine learning techniques. It is shown that the subsequent classification models are highly comparable. Finally the merits of a combined multivariate statistical and wavelet decomposition approach is considered. Techniques are applied to output signals from an experimental compressor rig.

Original languageEnglish
Title of host publicationProceedings of 2016 7th International Conference on Mechanical and Aerospace Engineering, ICMAE 2016
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages196-201
Number of pages6
ISBN (Electronic)9781467388290
DOIs
Publication statusPublished - 23 Aug 2016
Event7th International Conference on Mechanical and Aerospace Engineering - London, United Kingdom
Duration: 18 Jul 201620 Jul 2016
Conference number: 7

Conference

Conference7th International Conference on Mechanical and Aerospace Engineering
Abbreviated titleICMAE 2016
CountryUnited Kingdom
CityLondon
Period18/07/1620/07/16

Fingerprint

Wavelet decomposition
Compressors
Learning systems
Data reduction
Statistical methods
Health
Monitoring
Big data

Cite this

Smith, A., Gu, F., & Ball, A. (2016). Maintaining model efficiency, avoiding bias and reducing input parameter volume in compressor fault classification. In Proceedings of 2016 7th International Conference on Mechanical and Aerospace Engineering, ICMAE 2016 (pp. 196-201). [7549534] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICMAE.2016.7549534
Smith, Ann ; Gu, Fengshou ; Ball, Andrew. / Maintaining model efficiency, avoiding bias and reducing input parameter volume in compressor fault classification. Proceedings of 2016 7th International Conference on Mechanical and Aerospace Engineering, ICMAE 2016. Institute of Electrical and Electronics Engineers Inc., 2016. pp. 196-201
@inproceedings{595927b179474e36ad25178091070bcf,
title = "Maintaining model efficiency, avoiding bias and reducing input parameter volume in compressor fault classification",
abstract = "With the exponential growth in data collection and storage and the necessity for timely prognostic health monitoring of industrial processes traditional methods of data analysis are becoming redundant. Big data sets and huge volumes of inputs give rise to equally massive computational requirements. In this paper the differences in input parameter selection using a subset of the original variables and using data reduction techniques are compared. Each selection procedure being illustrated by both statistical methods and machine learning techniques. It is shown that the subsequent classification models are highly comparable. Finally the merits of a combined multivariate statistical and wavelet decomposition approach is considered. Techniques are applied to output signals from an experimental compressor rig.",
keywords = "big data, data compression, fault classification, MSPCA",
author = "Ann Smith and Fengshou Gu and Andrew Ball",
year = "2016",
month = "8",
day = "23",
doi = "10.1109/ICMAE.2016.7549534",
language = "English",
pages = "196--201",
booktitle = "Proceedings of 2016 7th International Conference on Mechanical and Aerospace Engineering, ICMAE 2016",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

Smith, A, Gu, F & Ball, A 2016, Maintaining model efficiency, avoiding bias and reducing input parameter volume in compressor fault classification. in Proceedings of 2016 7th International Conference on Mechanical and Aerospace Engineering, ICMAE 2016., 7549534, Institute of Electrical and Electronics Engineers Inc., pp. 196-201, 7th International Conference on Mechanical and Aerospace Engineering, London, United Kingdom, 18/07/16. https://doi.org/10.1109/ICMAE.2016.7549534

Maintaining model efficiency, avoiding bias and reducing input parameter volume in compressor fault classification. / Smith, Ann; Gu, Fengshou; Ball, Andrew.

Proceedings of 2016 7th International Conference on Mechanical and Aerospace Engineering, ICMAE 2016. Institute of Electrical and Electronics Engineers Inc., 2016. p. 196-201 7549534.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Maintaining model efficiency, avoiding bias and reducing input parameter volume in compressor fault classification

AU - Smith, Ann

AU - Gu, Fengshou

AU - Ball, Andrew

PY - 2016/8/23

Y1 - 2016/8/23

N2 - With the exponential growth in data collection and storage and the necessity for timely prognostic health monitoring of industrial processes traditional methods of data analysis are becoming redundant. Big data sets and huge volumes of inputs give rise to equally massive computational requirements. In this paper the differences in input parameter selection using a subset of the original variables and using data reduction techniques are compared. Each selection procedure being illustrated by both statistical methods and machine learning techniques. It is shown that the subsequent classification models are highly comparable. Finally the merits of a combined multivariate statistical and wavelet decomposition approach is considered. Techniques are applied to output signals from an experimental compressor rig.

AB - With the exponential growth in data collection and storage and the necessity for timely prognostic health monitoring of industrial processes traditional methods of data analysis are becoming redundant. Big data sets and huge volumes of inputs give rise to equally massive computational requirements. In this paper the differences in input parameter selection using a subset of the original variables and using data reduction techniques are compared. Each selection procedure being illustrated by both statistical methods and machine learning techniques. It is shown that the subsequent classification models are highly comparable. Finally the merits of a combined multivariate statistical and wavelet decomposition approach is considered. Techniques are applied to output signals from an experimental compressor rig.

KW - big data

KW - data compression

KW - fault classification

KW - MSPCA

UR - http://www.scopus.com/inward/record.url?scp=84987995240&partnerID=8YFLogxK

U2 - 10.1109/ICMAE.2016.7549534

DO - 10.1109/ICMAE.2016.7549534

M3 - Conference contribution

SP - 196

EP - 201

BT - Proceedings of 2016 7th International Conference on Mechanical and Aerospace Engineering, ICMAE 2016

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Smith A, Gu F, Ball A. Maintaining model efficiency, avoiding bias and reducing input parameter volume in compressor fault classification. In Proceedings of 2016 7th International Conference on Mechanical and Aerospace Engineering, ICMAE 2016. Institute of Electrical and Electronics Engineers Inc. 2016. p. 196-201. 7549534 https://doi.org/10.1109/ICMAE.2016.7549534