PSCMAP: A new tool for plant-wide oscillation detection

doi:10.1016/j.jprocont.2005.01.005

Journal of Process Control

Volume 15, Issue 8, December 2005, Pages 931-941

https://doi.org/10.1016/j.jprocont.2005.01.005 Get rights and content

Abstract

A commonly encountered issue in process industry is concerned with the detection of plant-wide oscillations. In this paper, a new visualization tool termed as the power spectral correlation map (PSCMAP) is proposed for this purpose. The proposed colour map is based on a new measure defined as the power spectral correlation index (PSCI). A simple clustering algorithm is developed to group blocks of variables with similar spectral shapes. The combined visualization tool is shown to be simple, effective and powerful in collecting variables with common frequency-domain behaviour in a multivariate process. The potential of the combined technique is illustrated by an application to two industrial processes, (i) a simulated pulp and paper process and (ii) a SE Asia refinery.

Section snippets

Motivation

The detection of plant-wide oscillations is concerned with the identification of a set of signals (or measurements) oscillating at similar frequencies, but not necessarily in phase with each other. Such a problem is of great practical significance since it involves the identification of common cause disturbances whose effects propagate to many units and thus may impact overall process performance. The causes of oscillations, their influence on poor performance, and their benefits to the economy

Power spectral correlation index

The power spectral correlation index (PSCI) is defined as the correlation between the power spectra of two different measurements. It is a measure of the similarity of spectral shapes, i.e., measure of the commonness of frequencies of oscillations. The procedure to calculate the correlation is illustrated in the block diagram shown in Fig. 1.

The DFTs that are used to calculate the spectrum are calculated after removal of means from the time-series data. However, the correlation used in the

Power spectral correlation map

For multivariate processes, the PSCI is a matrix of size m × m, where m is the number of measured variables. In order to provide an effective interpretation of the PSCI, the matrix is plotted as a colour map, which is termed as the power spectral correlation map. The intensity as well as the type of colour in the map is assigned in proportion to the value of the correlation index. This mapping is performed according to the choice of colour and the number of shades in that colour.

An important

Case study 1: Entech data

We first consider the data set from a simulated industrial process, courtesy of Entech Control Inc.

The simulated process shown in Fig. 2 consists of a pulp manufacturing process, where the hardwood and softwood pulps are mixed to give a stream of desired composition. The data set comprised 1934 samples from 12 process measurements (tags), each of which was associated with 12 control loops. The objective of this analysis is to detect oscillations in the loops and isolate those loops with common

Conclusions

A new visualization tool, namely, the PSCMAP has been proposed to identify and extract the measurements from a large plant with several variables oscillating at similar frequencies. The visualization tool is based on a new measure, PSCI, which is a non-centered correlation between the spectra of measurements. The proposed method requires minimal computational complexity and effort. A simple clustering algorithm has been introduced to automate the re-arrangement of variables with similar

Acknowledgement

The authors are grateful for the financial support of the Natural Science and Engineering Research Council (NSERC, Canada), Matrikon (Edmonton, Alberta) and the Alberta Science and Research Authority (ASRA) through the NSERC-Matrikon-ASRA Industrial Research Chair in Process Control.

References (11)

N.F. Thornhill et al.
Detection and diagnosis of oscillation in control loops
Contr. Eng. Pract.
(1997)
N.F. Thornhill et al.
Spectral principal component analysis of dynamic process data
Contr. Eng. Pract.
(2002)
N.F. Thornhill et al.
Detection of multiple oscillations in control loops
J. Process Contr.
(2003)
S.J. Qin
Control performance monitoring—a review and assessment
Comput. Chem. Eng.
(1998)
L. Desborough, R. Miller, Increasing customer value of industrial performance monitoring—Honeywell’s experience,...

There are more references available in the full text version of this article.

Cited by (42)

Complex probabilistic slow feature extraction with applications in process data analytics
2021, Computers and Chemical Engineering
Citation Excerpt :
The identification of oscillations using data visualization is impractical due to the presence of various frequency segments coupled with measurement noise. Hence, the issue of plant-wide oscillations diagnosis has been studied extensively using the non-linearity tests (Choudhury et al., 2006), spectral envelope methods (Jiang et al., 2007; Tangirala et al., 2005), and process topology-based methods (Kaminski and Blinowska, 1991; Baccalá and Sameshima, 2001; Marinazzo et al., 2008; Tangirala et al., 2007; Kannan and Tangirala, 2014; Sugihara et al., 2012; Krishna and Tangirala, 2019). In this article, we propose the complex probabilistic slow feature analysis (CPSFA) model to extract oscillating features in the presence of noise.
Today, in modern industrial processes, thousands of correlated process variables are measured and stored. Dimension reduction techniques are often employed to construct informative features by discarding redundant information. Slow feature analysis is one such technique that extracts the slowly varying patterns from measured data. Oscillatory behaviour is prevalent in process data due to inadequate control loop tuning and external disturbances such as diurnal temperature variation. Extracting these oscillatory patterns is vital in applications such as control loop monitoring, fault diagnosis. Slow feature analysis may not extract oscillating patterns when the signal to noise ratio is low in process data. This paper proposes the complex probabilistic formulation that extracts slow oscillatory features. We also present the Expectation-Maximization algorithm to obtain the optimal parameter estimates. Finally, three case studies are presented to illustrate the efficacy of the proposed formulation in soft sensing and fault detection applications.
Detection and root cause analysis of multiple plant-wide oscillations using multivariate nonlinear chirp mode decomposition and multivariate Granger causality
2021, Computers and Chemical Engineering
Citation Excerpt :
Researchers tried to solve this problem with multivariate processing technology. Spectral principal component analysis (Thornhill et al., 2002), power spectral correlation map (Tangirala et al., 2005), spectral envelope (Jiang et al., 2007), multivariate spectral decomposition (El-Ferik et al., 2012) are utilized in succession. However, these Fourier-based methods require the collected signals to be stationary, which limits the adaptability and practicability.
Plant-wide oscillation detection and root cause diagnosis are important for maintaining control performance. Existing methods are mainly limited to detecting single and time-invariant plant-wide oscillations. In this paper, a data-driven model combining multivariate nonlinear chirp mode decomposition (MNCMD) with multivariate Granger causality (MGC) is proposed to detect and analyze root causes for multiple plant-wide oscillations in process control system. First, an MNCMD-based detector is developed to capture the multiple plant-wide oscillations, where oscillating variables caused by different sources are automatically clustered into various groups. Then, MGC is applied to each group to obtain the root causes of multiple plant-wide oscillations. Compared with state-of-the-art detection methods, the proposed approach shows better performance in the following aspects: (i) ability to extract both single/multiple plant-wide oscillations; (ii) capability to process both time-invariant/time-varying oscillations and provide accurate time-frequency information. This work also outperforms original Granger causality and nonlinearity index-based method in providing clearer causal network. The effectiveness and advantages of the proposed approach are demonstrated with the help of both simulation and industrial case studies.
An integrated application of control performance assessment and root cause analysis in refinery control loops
2020, IFAC-PapersOnLine
Assessing the performance of control loops is an important component of Control Performance Monitoring (CPM) systems. Most of the industrial chemical processes have a large number of control loops interacting with each other in a complex way due to material and energy integration in the plant. A problem occurring in a certain control loop can easily upset the performance of the other control loops. Therefore, identification of the ”bad” control loops causing a plant-wide disturbances is a crucial task. In this work, an integrated approach covering performance assessment and interaction analysis is proposed to detect the ”bad” loops based on their performances. First, Minimum Variance Control (MVC) benchmark is used to screen-out the poor performing loops. Then, the spectral envelope method utilizing frequency analysis is used to identify the common oscillation periods among the loops under study. Finally, Granger causality is used to plot the interaction map between the loops. even though these methods are well developed and used for several purposes separately, we present an integrated approach which focuses and analyzes the ”bad loops”. The developed approach has been tested in a refinery plant having 18 control loops. The results show that the proposed method is clearly able to identify and isolate the root-cause control loops. The validation of results and further improvements in the control loops under study have been given.
Oscillation detection in process industries – Part I: Review of the detection methods
2019, Journal of Process Control
Citation Excerpt :
Different approaches are presented in the third group. The technique proposed by Tangirala et al. [57] is a powerful and easy-to-implement tool. Its application is recommended as the first step on plant-wide oscillation diagnosis.
Oscillatory control loop is a frequent problem in process industries. Its incidence reduces product uniformity and increases both energy consumption and raw material waste. These impacts strongly reduce plant profitability, making fundamental the removal of such oscillation. The first step in oscillation removal is its detection, which should preferably be performed automatically to enable the evaluation of the entire plant. A large number of these automatic techniques have been proposed over the last years. Thus, understanding and selecting a proper technique for the desired application is very time-consuming. This work is the first of a two-part study on oscillation detection in process industries. While the second part focuses on the evaluation of the detection techniques applied to industrial data, this part mainly includes an extensive review of the works on the topic and discussions on the definition, types, strength, and causes of oscillation. Also, the techniques are clustered into groups, and their pros and cons are confronted.
Application of correlation analysis techniques to surface wave testing for the evaluation of reinforced concrete structural elements
2019, NDT and E International
An experimental study using surface wave testing to monitor the progression of fatigue damage in five natural scale reinforced concrete beams is presented. The work presented innovates in the manner the results from the surface wave testing is analyzed. Instead of evaluating the velocity and attenuation of the surface waves, the determination of the damage results from the correlation analysis between the recorded signals. Three correlation analysis techniques were assessed: the Pearson correlation coefficient, the power spectral correlation index and the coherence spectrum. From the results obtained, the coherence spectrum showed potential to monitor the progression of incremental damage.
Improved oscillation detection via noise-assisted data analysis
2018, Control Engineering Practice
Oscillation detection is usually a precursor to more advanced performance monitoring steps such as plant wide oscillation detection and root cause detection. Therefore any false or missed detection can have serious implications. Oscillation detection is a challenging problem due to the presence of noise and multiple modes in the plant data. This paper presents an improved and robust automatic oscillation detection algorithm based on noise-assisted data analysis that can handle multiple oscillatory modes in the presence of both coloured and white noise along with non-stationary effects. The dyadic filter bank property of multivariate empirical mode decomposition has been used to accurately detect the oscillations and to calculate the associated characteristics. This work improves upon the existing auto covariance function based methods. The robustness and reliability of the proposed scheme is demonstrated via simulation and industrial case studies.

View all citing articles on Scopus

View full text

Short communicationPSCMAP: A new tool for plant-wide oscillation detection

Abstract

Section snippets

Motivation

Power spectral correlation index

Power spectral correlation map

Case study 1: Entech data

Conclusions

Acknowledgement

Contr. Eng. Pract.

Contr. Eng. Pract.

J. Process Contr.

Control performance monitoring—a review and assessment

Comput. Chem. Eng.

Short communication
PSCMAP: A new tool for plant-wide oscillation detection