ERA

Download the full-sized PDF of Simultaneous Gross Error Detection and Data Reconciliation Using Gaussian Mixture DistributionDownload the full-sized PDF

Analytics

Share

Permanent link (DOI): https://doi.org/10.7939/R34747543

Download

Export to: EndNote  |  Zotero  |  Mendeley

Communities

This file is in the following communities:

Graduate Studies and Research, Faculty of

Collections

This file is in the following collections:

Theses and Dissertations

Simultaneous Gross Error Detection and Data Reconciliation Using Gaussian Mixture Distribution Open Access

Descriptions

Other title
Subject/Keyword
Data Reconciliation
Data Rectification
Gross Error Detection
Type of item
Thesis
Degree grantor
University of Alberta
Author or creator
Alighardashi, Hashem
Supervisor and department
Huang, Biao (Department of Chemical and Materials Engineering)
Examining committee member and department
Huang, Biao (Department of Chemical and Materials Engineering)
Prasad, Vinay (Department of Chemical and Materials Engineering)
Koch, Bob (Department of Mechanical Engineering)
Department
Department of Chemical and Materials Engineering
Specialization
Process Control
Date accepted
2017-09-27T16:11:45Z
Graduation date
2017-11:Fall 2017
Degree
Master of Science
Degree level
Master's
Abstract
The intensive competitive nature of the world market, the growing significance of quality products, and the increasing importance and the number of safety and environmental issues and regulations, respectively, have increased the need for fast and low-cost changes in chemical processes to enhance their performance. Any possible changes and modifications in a system in order to control, optimize, evaluate the behavior of the process, or achieve the maximal performance of the system require clear understanding and knowledge of its actual state. This information is obtained by processing a data set - collecting it, ameliorating its accuracy, and storing/using it for further analysis. It should be emphasized that in today’s highly competitive world market, increasing the accuracy of measurements by resolving even small errors can result in substantial improvements in plant efficiency and economy. Industrial process measurements play a significant role in online optimization, process monitoring, identification, and control. These measurements are used to make decisions which potentially influence product quality, plant safety, and profitability. Nonetheless, they are inherently contaminated by errors, which may be random and/or systematic/gross errors, due to sensor accuracy, improper instrumentation, poor calibration, process leak, and so on. The objective of data reconciliation and gross error detection is the estimation of the true states and the detection of any faults in the instruments which could seriously degrade the performance of the system. Data reconciliation techniques deal with the problem of improving the accuracy of raw process measurements and their application allows optimal adjustment of measurement values to satisfy material and energy constraints. These methods also make possible estimation of the unmeasured variables. However, data reconciliation approaches do not always provide valid estimates of the actual states, and the presence of gross errors in the measurements significantly affect the accuracy levels that can be accomplished using reconciliation. Therefore, the main focus of this work is to develop a framework to obtain the accurate estimates of reconciled values while reducing the impact of gross errors. In reality, operating conditions under which a process works change with different circumstances. Therefore, it is vital to develop a model that is capable of identifying and switching between operating regions. To this end, a method is proposed for simultaneous gross error detection and rectification of a data set which contains different operating regions. First, the data set is divided into several clusters based on the number of operating regions. Then, the same operation, i.e., data rectification is performed on each operating region. It must be noted that all of the proposed approaches in this thesis do not require to preset the parameters of the error distribution model, rather they are determined as part of the solution. They are also applicable to problems with both linear and nonlinear constraints, in addition to the ability to determine the magnitude of gross errors. Furthermore, these methods/approaches detect partial gross errors, so it is not required to assume that gross errors exist in the entire data set. Finally, the performance of the proposed methods is verified through various simulation studies and realistic examples.
Language
English
DOI
doi:10.7939/R34747543
Rights
This thesis is made available by the University of Alberta Libraries with permission of the copyright owner solely for the purpose of private, scholarly or scientific research. This thesis, or any portion thereof, may not otherwise be copied or reproduced without the written consent of the copyright owner, except to the extent permitted by Canadian copyright law.
Citation for previous publication

File Details

Date Uploaded
Date Modified
2017-09-27T22:11:46.235+00:00
Audit Status
Audits have not yet been run on this file.
Characterization
File format: pdf (Portable Document Format)
Mime type: application/pdf
File size: 1353445
Last modified: 2017:11:08 16:35:24-07:00
Filename: Alighardashi_Hashem_201709_Master.pdf
Original checksum: a67f54518881c2713f15a0f9833591d2
Well formed: true
Valid: true
Page count: 130
Activity of users you follow
User Activity Date