Intelligent process monitoring by interfacing knowledge-based systems and multivariate statistical monitoring

Norvilas, Aras; Negiz, Antoine; DeCicco, Jeffrey; Çınar, Ali

doi:10.1016/s0959-1524(99)00057-8

Cited by 66 publications

(39 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Vedam and Venkatasubramanian (1999) and Chiang and Braatz (2003) showed enhanced diagnosis using measurement-based analysis if a qualitative model is available. Norvilas et al (2000) combined multivariate statistical analysis and an expert system for the same purpose, while Leung and Romagnoli (2002) integrated a multivariate statistical analysis method with cause and effect map of a process, which was set up manually, to help in the diagnosis of faults. Lee et al, (2003) also combined SDGs with multivariate statistical analysis for enhanced diagnosis.…”

Section: Combination Methodsmentioning

confidence: 99%

Cause-and-effect analysis in chemical processes utilizing XML, plant connectivity and quantitative process history

Thambirajah

Benabbas

Bauer

et al. 2009

Computers & Chemical Engineering

View full text Add to dashboard Cite

Disturbances that spread plant-wide in a chemical process pose challenges to maintenance staff. Connections within the plant and the presence of multiple causal paths mean it is not straightforward to locate the root disturbance because the effects can propagate and be detected elsewhere. Measurementbased methods use quantitative process history to generate hypotheses about the root cause, while a separate strand of work in the literature has used causal maps and digraphs. It has been reported that both approaches can give spurious solutions, however. The idea behind this article is to reduce the number of spurious solutions by combining basic and readily-available information about the connectivity of the process with the results from causal measurement-based analysis. Connectivity information is captured from an XML description of the process schematic that complies with the CAEX schema. The capabilities of the approach and its potential for future development are discussed.

show abstract

Section: Combination Methodsmentioning

confidence: 99%

Cause-and-effect analysis in chemical processes utilizing XML, plant connectivity and quantitative process history

Thambirajah

Benabbas

Bauer

et al. 2009

Computers & Chemical Engineering

View full text Add to dashboard Cite

show abstract

“…As such, FDD should also involve the use of the process knowledge. This constitutes a group of methods referred to as combined data driven and knowledge based FDD [15][16][17]33] . An example is the well-known KBS technique, which uses historical operational data and understanding and the rules of the process to perform the required FDD.…”

Section: Knowledge Based Fdd As One Of Active Aspects For Data Drivenmentioning

confidence: 99%

Data Driven Fault Diagnosis and Fault Tolerant Control: Some Advances and Possible New Directions

Wang¹,

Chai²,

Ding³

et al. 2009

Acta Automatica Sinica

View full text Add to dashboard Cite

This paper presents a selected survey covering the advances of fault diagnosis and fault tolerant control using data driven techniques. A brief summary of the general developments in fault detection and diagnosis for industrial processes is given, which is then followed by discussions on the widely used data driven and knowledge-based techniques. A successful application example is also given, which deals with faults caused by the misplacement of control loop set points and several areas of potential future directions are included in the paper. Key words Fault detection, fault diagnosis, fault tolerant control, data driven techniques Fault detection and diagnosis (FDD) and fault tolerant control (FTC) have been the subject of considerable interest in the control research community [1−39] . This is in response to the ever increasing requirements on the reliable operation of control systems, which are, in most cases, subject to a number of faults either in the internal closed loops or from environmental factors. Once system faults have occurred, they can cause unrecoverable losses and result in unacceptable environmental pollution, etc. Occasionally, the occurrence of a minor fault has resulted in disastrous effects. For example, it has been observed that faults have caused a 3 % ∼ 8 % reduction of the oil production in the United States, leading to $ 20 billion losses in the country s economy per year. Also, in 1997, the faults in a chemical plant in Beijing caused heavy direct losses. Therefore, effective FDD is of vital importance to the safe operation of industrial plants. Indeed, FDD and FTC have now become an integral part of industrial process control.In general, system faults can be grouped into several categories, namely, actuator faults, sensor faults, system faults and also abnormal operating faults caused by either the misplacement of control loop set points or unexpected variations in the raw materials to be processed. The purpose of FDD is to use available signals to detect, identify, and isolate possible sensor faults, actuator faults, and system faults. Conversely, FTC calculates the required actions (either controller modification or reconfiguration) so that the system can still continue to operate safely even under faulty conditions [2][3]40] . In terms of condition monitoring or FDD, the existing methods can also be grouped into the following two categories: 1) Model based FDD; 2) Data driven FDD including knowledge based FDD. In the early days (1980 s onwards), model based FDD constituted the main stream of research, and a number of techniques were developed. Depending on whether the system model can be represented as either a state space model or an input-output model, FDD can be classified into two groups: observer based FDD [1] and system identification based FDD [4] . Also, to combine the best features of these two approaches, there is another group of FDD methods called adaptive observer based fault diagnosis, which uses parameter tuning principles from model reference adap- tive ...

show abstract

“…In terms of training sets used the novelty detectors are either restricted (shown as "R") or broad (shown as "B") covering selected machining conditions where the value/s shown inside ( ) represent the relevant depth of cut associated with training data. Abbreviations used to describe tested conditions are: "c" for depth of cut in mm, "1-t" for one broken tooth, "2-t" for two broken teeth, "N" for normal tool conditions at specified machining parameters, and "Nx" for selected normal tool conditions at specific machining parameters that were deliberately excluded from the range of machining setting values used during training (7,9) 8, 10,11,13,14,16,17,19,20,22,23, & 25 B(5,7,9) 1-7, 8,10,11,13,14,16,17,19,20,22,23,& 25 of tool condition thus allowed the accuracy of NDCs' decisions to be examined and a percentage value given for each group of time-series windows that was tested by each NDC. Tables 3 and 4 describe the various machining conditions that were used to train the NDCs.…”

Section: The Response Variablementioning

confidence: 99%

“…Recent developments in MSPC using artificial intelligence include intelligent process monitoring by interfacing multivariate statistical process monitoring techniques and knowledge-based systems [9], and extension of the basic principle of partial least squares from linear to non-linear domain through the application of well known NN architectures, e.g. the radial basis functions (RBFs) [10].…”

Section: Introductionmentioning

confidence: 99%

Novelty detection for practical pattern recognition in condition monitoring of multivariate processes: a case study

Zorriassatine

Al‐Habaibeh

Parkin

et al. 2005

Int J Adv Manuf Technol

View full text Add to dashboard Cite

Multivariate statistical process control (MSPC) can be applied for condition monitoring (CM) purposes. MSPC is implemented using a variety of techniques including neural networks (NNs). In situations when the number of process attributes is sufficiently large (e.g. 10 or more) concerns can arise with respect to training of NNs for pattern recognition. A classification method known as novelty detection (ND) can provide an effective alternative to conventional NN solutions that suffer from the above problem. Despite its great potential, ND is still unknown to the broad community of manufacturing engineers. This paper successfully demonstrates the ability of ND, using Gaussian mixture models, to notify operators of an end-milling process of the presence of faulty tool conditions. A significant achievement is that ND is used to identify abnormal time-series patterns as opposed to individual vectors of multiple simultaneous measurements related to abnormal conditions. Such patterns are found in windowed streams of signals related to 10 different process features (with an effective problem dimension of 140). The paper also investigates some of the issues related to implementation of ND for pattern recognition in condition monitoring.

show abstract

Intelligent process monitoring by interfacing knowledge-based systems and multivariate statistical monitoring

Cited by 66 publications

References 12 publications

Cause-and-effect analysis in chemical processes utilizing XML, plant connectivity and quantitative process history

Cause-and-effect analysis in chemical processes utilizing XML, plant connectivity and quantitative process history

Data Driven Fault Diagnosis and Fault Tolerant Control: Some Advances and Possible New Directions

Novelty detection for practical pattern recognition in condition monitoring of multivariate processes: a case study

Contact Info

Product

Resources

About