Application and Simulation Study of the Hotelling's T2 Control Chart to Monitor a Wastewater Treatment Process.

Saved in:

Bibliographic Details
Title:	Application and Simulation Study of the Hotelling's T² Control Chart to Monitor a Wastewater Treatment Process.
Authors:	Capilla, C.¹
Source:	Environmental Engineering Science. Feb2009, Vol. 26 Issue 2, p333-342. 10p. 1 Chart, 10 Graphs.
Subjects:	Wastewater treatment, Waste management, Sewage purification, Sewage disposal plants, Water quality, Water quality management, Monte Carlo method
Abstract:	The purpose of this paper is to study the application of the Hotelling's T2 statistic to assess the performance of a wastewater treatment process. Standard control charts are widely used in engineering for quality improvement through the detection of special causes of variability. These methods assume that the observations are independent. In wastewater treatment processes the data are frequently autocorrelated. One possible way to apply control charts to autocorrelated observations is to monitor the one-step ahead forecast errors after estimating the process time series model. In this paper the application of control charts to individual forecast errors of the multivariate time series model from a wastewater process is discussed. Hotelling's T2 statistics are computed using the sample covariance matrix, the successive differences estimator and the minimum volume ellipsoid estimator. A Monte Carlo simulation study shows that the application of these charts to the time series model residuals is quite effective in detecting the mean changes. However, the multivariate T2 charts are more powerful when applied to independent observations. The simulation study also indicates that the autocorrelation affects the properties of the T2 control chart giving a large false alarm rate when using the successive difference estimator, or affecting its power to detect level shifts when using the sample covariance matrix or the minimum volume ellipsoid estimator. These two charts have essentially no power to detect the simulated shifts. [ABSTRACT FROM AUTHOR]
	Copyright of Environmental Engineering Science is the property of Mary Ann Liebert, Inc. and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database:	Engineering Source

Description
Abstract:	The purpose of this paper is to study the application of the Hotelling's T2 statistic to assess the performance of a wastewater treatment process. Standard control charts are widely used in engineering for quality improvement through the detection of special causes of variability. These methods assume that the observations are independent. In wastewater treatment processes the data are frequently autocorrelated. One possible way to apply control charts to autocorrelated observations is to monitor the one-step ahead forecast errors after estimating the process time series model. In this paper the application of control charts to individual forecast errors of the multivariate time series model from a wastewater process is discussed. Hotelling's T2 statistics are computed using the sample covariance matrix, the successive differences estimator and the minimum volume ellipsoid estimator. A Monte Carlo simulation study shows that the application of these charts to the time series model residuals is quite effective in detecting the mean changes. However, the multivariate T2 charts are more powerful when applied to independent observations. The simulation study also indicates that the autocorrelation affects the properties of the T2 control chart giving a large false alarm rate when using the successive difference estimator, or affecting its power to detect level shifts when using the sample covariance matrix or the minimum volume ellipsoid estimator. These two charts have essentially no power to detect the simulated shifts. [ABSTRACT FROM AUTHOR]
ISSN:	10928758
DOI:	10.1089/ees.2007.0358