ArticlePublisher preview available

Causal network construction based on KICA-ECCM for root cause diagnosis of industrial processes

Authors:
Yayin He
Yayin He
Yayin He
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Xiangshun Li
Xiangshun Li
Xiangshun Li
  • This person is not on ResearchGate, or hasn't claimed this research yet.
Read publisher preview
Download citation
To read the full-text of this research, you can request a copy directly from the authors.

Abstract and Figures

Root cause diagnosis is able to find the propagation path of faults timely when the fault occurs. Therefore, it is of key significance in the maintenance and fault diagnosis of industrial systems. A commonly used method for root cause diagnosis is causal analysis method. In this work, a causal analysis method Extended Convergent Cross Mapping (ECCM) algorithm is used for root cause diagnosis of industry, however, it has difficulties in dealing with large amounts of steady state data and obtaining accurate propagation paths. Therefore, a causal analysis method based on Kernel Independent Component Analysis (KICA) and ECCM is proposed in this study to deal with the above problems. First, the KICA algorithm is used to detect faults to get the transition process data. Second, the ECCM algorithm is used to construct causal relationship among variables based on the transition process data to construct the fault propagation path diagram. Finally, the effectiveness of the proposed KICA-ECCM algorithm is tested by using the Tennessee Eastman Process and Industrial Process Control Test Facility platform. Compared with the ECCM and GC algorithm, the KICA-ECCM algorithm performs better in terms of accuracy and efficiency.
Figures - available from: Cluster Computing
This content is subject to copyright. Terms and conditions apply.
A preview of this full-text is provided by Springer Nature.
Learn more
Content available from Cluster Computing
This content is subject to copyright. Terms and conditions apply.
Causal network construction based on KICA-ECCM for root cause
diagnosis of industrial processes
Yayin He
1
•Xiangshun Li
1
Received: 5 April 2024 / Revised: 18 June 2024 / Accepted: 5 July 2024 / Published online: 20 July 2024
ÓThe Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2024, corrected publication 2024
Abstract
Root cause diagnosis is able to find the propagation path of faults timely when the fault occurs. Therefore, it is of key
significance in the maintenance and fault diagnosis of industrial systems. A commonly used method for root cause
diagnosis is causal analysis method. In this work, a causal analysis method Extended Convergent Cross Mapping (ECCM)
algorithm is used for root cause diagnosis of industry, however, it has difficulties in dealing with large amounts of steady
state data and obtaining accurate propagation paths. Therefore, a causal analysis method based on Kernel Independent
Component Analysis (KICA) and ECCM is proposed in this study to deal with the above problems. First, the KICA
algorithm is used to detect faults to get the transition process data. Second, the ECCM algorithm is used to construct causal
relationship among variables based on the transition process data to construct the fault propagation path diagram. Finally,
the effectiveness of the proposed KICA-ECCM algorithm is tested by using the Tennessee Eastman Process and Industrial
Process Control Test Facility platform. Compared with the ECCM and GC algorithm, the KICA-ECCM algorithm per-
forms better in terms of accuracy and efficiency.
Keywords Root cause diagnosis Kernel independent component analysis (KICA) Extended convergent cross mapping
(ECCM)
1 Introduction
In complex industrial systems, it is particularly important to
detect the occurrence of faults. In recent years, due to the rapid
development of big data technology, data-driven multivariate
process monitoring methods have been widely applied.
The multivariate process monitoring methods like
Principal Component Analysis(PCA) and Independent
Component Analysis (ICA) have been applied in industry
process very early [1–4]. However, the PCA and ICA
algorithm have limitations when processing data with
nonlinear structures. In order to solve the problems caused
by nonlinear data, the Kernel Principal Component
Analysis(KPCA) and Kernel Independent Component
Analysis(KICA) nonlinear process monitoring technology
are proposed [5–7], which have also been widely used.
However, the purpose of fault detection is to monitor
whether the process is functioning correctly, no further
analysis of the fault is performed and the fault propagation
paths and diagnosis the root fault cause is often neglected.
Root cause diagnosis refers to the systematic methods and
technologies used to trace the root cause variables of failures
in industrial systems. After identifying the root variables that
cause the fault, personnel can timely address them, which is
significantly important for the reliability and safety of the
system. Therefore, the study of root cause diagnosis has
received significant attention in recent years. As a method for
root cause diagnosis, causal analysis methods can be highly
effective in identifying the root causes of faults and corre-
sponding propagation pathways, which are commonly used in
the field for root cause diagnosis [8,9].
Causal analysis methods can be divided into two cate-
gories: knowledge-based methods and data-based methods.
Knowledge-based methods such as Failure Modes and Effects
&Xiangshun Li
lixiangshun@whut.edu.cn
Yayin He
heyayin@whut.edu.cn
1
School of Automation, Wuhan University of Technology,
Wuhan 430079, China
123
Cluster Computing (2024) 27:11891–11909
https://doi.org/10.1007/s10586-024-04663-5(0123456789().,-volV)(0123456789().,-volV)
Content courtesy of Springer Nature, terms of use apply. Rights reserved.
ResearchGate has not been able to resolve any citations for this publication.
Inferring a Causal Relationship between Environmental Factors and Respiratory Infections Using Convergent Cross-Mapping
Article
Full-text available
  • May 2023
  • Entropy
The incidence of respiratory infections in the population is related to many factors, among which environmental factors such as air quality, temperature, and humidity have attracted much attention. In particular, air pollution has caused widespread discomfort and concern in developing countries. Although the correlation between respiratory infections and air pollution is well known, establishing causality between them remains elusive. In this study, by conducting theoretical analysis, we updated the procedure of performing the extended convergent cross-mapping (CCM, a method of causal inference) to infer the causality between periodic variables. Consistently, we validated this new procedure on the synthetic data generated by a mathematical model. For real data in Shaanxi province of China in the period of 1 January 2010 to 15 November 2016, we first confirmed that the refined method is applicable by investigating the periodicity of influenza-like illness cases, an air quality index, temperature, and humidity through wavelet analysis. We next illustrated that air quality (quantified by AQI), temperature, and humidity affect the daily influenza-like illness cases, and, in particular, the respiratory infection cases increased progressively with increased AQI with a time delay of 11 days.
View
Detecting time-delayed causal interaction between Northern Hemisphere annular mode and winter surface air temperature over Northeast China: a case study of 2009/2010 winter
Article
Full-text available
  • Nov 2021
  • THEOR APPL CLIMATOL
The extremely low value of the Northern Hemisphere annular mode (NAM) that occurred in the winter of 2009/2010 led to rare extreme low temperatures and heavy snow disasters over Northeast China. Studies on the effect of NAM on temperature often rely on correlation analysis, synthetic analysis, regression analysis, and other statistical methods. However, although the statistical method is instructive to the discussion of physical mechanisms, it cannot prove whether there is causality between variables. Here, we apply extended convergent cross mapping (ECCM) to detect the time-delayed causal interaction between NAM and winter surface air temperature (SAT) over Northeast China. Results indicate that the winter SAT information is encoded in the contemporaneous 1000 hPa NAM and vice versa, showing the contemporaneous bidirectional causality between them. In addition, we detected a unidirectional causality between 500 hPa NAM and winter SAT. NAM at 500 hPa can influence the SAT but not vice versa. Moreover, we examine the time-delayed causality between winter SAT and the stratospheric NAM at 50 hPa. Stratospheric NAM can influence the SAT approximately 10 days later. These causal relationships and feedback effects in dynamic systems have opened up the possibility of improving the prediction of winter SAT over Northeast China by using the NAM as an external factor.
View
Causal analysis of the temperature impact on deep-sea biodiversity
Article
Full-text available
The deep sea comprises more than 90% of the ocean; therefore, understanding the controlling factors of biodiversity in the deep sea is of great importance for predicting future changes in the functioning of the ocean system. Consensus has recently been increasing on two plausible factors that have often been discussed as the drivers of deep-sea species richness in the contexts of the species-energy and physiological tolerance hypotheses: (i) seafloor particulate organic carbon (POC) derived from primary production in the euphotic zone and (ii) temperature. Nonetheless, factors that drive deep-sea biodiversity are still actively debated potentially owing to a mirage of correlations (sign and magnitude are generally time dependent), which are often found in nonlinear, complex ecological systems, making the characterization of causalities difficult. Here, we tested the causal influences of POC flux and temperature on species richness using long-term palaeoecological datasets derived from sediment core samples and convergent cross mapping, a numerical method for characterizing causal relationships in complex systems. The results showed that temperature, but not POC flux, influenced species richness over 103-104-year time scales. The temperature-richness relationship in the deep sea suggests that human-induced future climate change may, under some conditions, affect deep-sea ecosystems through deep-water circulation changes rather than surface productivity changes.
View
Atmospheric blocking events in the North Atlantic: trends and links to climate anomalies and teleconnections
Article
Full-text available
  • Apr 2021
  • CLIM DYNAM
Atmospheric blockings are generally associated with large-scale high-pressure systems that interrupt west-to-east atmospheric flow in mid and high latitudes. Blockings cause several days of quasi-stationary weather conditions, and therefore can result in monthly or seasonal climate anomalies and extreme weather events on the affected regions. In this paper, the long-term coupled CERA-20C reanalysis data from 1901 to 2010 are used to evaluate the links between blocking events over the North Atlantic north of 35° N, and atmospheric and oceanic modes of climate variability on decadal time scales. This study indicates more frequent and longer lasting blocking events than previous studies using other reanalyses products. A strong relationship was found between North Atlantic blocking events and North Atlantic Oscillation (NAO), Atlantic Multidecadal Oscillation (AMO) and Baffin Island–West Atlantic (BWA) indices, in fall, winter and spring. More blocking events occur during the negative phases of the NAO index and positive phases of the BWA mode. In some situations, the BWA patterns provide clearer links with the North Atlantic blocking occurrence than with the NAO alone. The correlation between the synchronous occurrences of AMO and blocking is generally weak, although it does increase for a lag of about 6–10 years. Convergent cross mapping (CCM) furthermore demonstrates a significant two-way causal effect between blocking occurrences and the NAO and BWA indices. Finally, while we find no significant trends in blocking frequencies over the last 110 years in the Northern Hemisphere, these events become longer lasting in summer and fall, and more intense in spring in the North Atlantic.
View
Evidence for a Causal Relationship between the Solar Cycle and Locust Abundance
Article
Full-text available
  • Dec 2020
Citation: Cheke, R.A.; Young, S.; Wang, X.; Tratalos, J.A.; Tang, S.; Cressman, K. Evidence for a Causal Relationship between the Solar Cycle and Locust Abundance. Agronomy 2021, 11, 69. https://doi.org/10.3390/agronomy11010069.
View
Inferring species interactions using Granger causality and convergent cross mapping
Article
Full-text available
  • Mar 2021
  • THEOR ECOL-NETH
Identifying directed interactions between species from time series of their population densities has many uses in ecology. This key statistical task is equivalent to causal time series inference, which connects to the Granger causality (GC) concept: x causes y if x improves the prediction of y in a dynamic model. However, the entangled nature of nonlinear ecological systems has led to question the appropriateness of Granger causality, especially in its classical linear multivariate autoregressive (MAR) model form. Convergent cross mapping (CCM), a nonparametric method developed for deterministic dynamical systems, has been suggested as an alternative. Here, we show that linear GC and CCM are able to uncover interactions with surprisingly similar performance, for predator-prey cycles, 2-species deterministic (chaotic), or stochastic competition, as well as 10- and 20-species interaction networks. We found no correspondence between the degree of nonlinearity of the dynamics and which method performs best. Our results therefore imply that Granger causality, even in its linear MAR(p) formulation, is a valid method for inferring interactions in nonlinear ecological networks; using GC or CCM (or both) can instead be decided based on the aims and specifics of the analysis.
View
View
An improved dynamic latent variable regression model for fault diagnosis and causal analysis
Article
  • Nov 2022
The advancement of industrial techniques has imposed a high demand for powerful machine learning algorithms to model the increasingly complicated relations in the data. Among them, dynamic models are widely studied to capture the inevitable temporal relations. However, most existing methods only focus on the dynamics between input and output data, failing to exploit other valuable information of the output. In this article, an improved dynamic latent variable regression method is proposed to capture both auto‐correlations and cross‐correlations between input and output with an auto‐regressive exogenous model, which is referred to as DrLVR‐ARX. Further, a DrLVR‐ARX based fault detection and diagnosis framework is designed to identify the root causes of a detected fault. The framework systematically integrates reconstruction‐based contribution, time‐domain Granger causality, and conditional spectral Granger causality to determine and locate the assignable causes. The effectiveness of the proposed algorithms is demonstrated with two industrial processes.
View
Time-varying Granger causality tests in the energy markets: A study on the DCC-MGARCH Hong test
Article
  • Jun 2022
  • ENERG ECON
The analysis of causality among oil prices and, in general, between financial and economic variables is of central relevance in applied economic studies. The recent contribution of Lu et al. (2014) proposes a new causality test, the DCC-MGARCH Hong test. We show that the critical values of the test statistic should be evaluated through simulations to avoid potential Type I errors. We also note that rolling Hong tests represent a more viable solution in the presence of short-lived causality periods.
View
Causal network construction based on convergent cross mapping (CCM) for alarm system root cause tracing of nonlinear industrial process
Article
  • Jan 2020
In terms of an alarm system, the propagation of a fault is identified as the main reason for low efficiency and the leading cause of dramatic industrial accidents. Thus, tracing the root causes of faulty conditions that lead to alarm floods is necessary. For root cause tracing, a widely accepted method is to characterize the process by causality at first and then trace the root causes. This work focuses on the former part. The conventional techniques to deal with causal analysis of industrial processes have difficulty in handling the nonlinearity of variables, obtaining accurate probability density and time lag, etc. In this work, a novel causal network construction method based on convergent cross mapping (CCM) that accurately describe process causality was proposed to deal with the above problems. First, the original monitoring variables were determined by a maximum Lyapunov method to determine whether they were chaotic time series, which aims to judge whether the application conditions of CCM can be satisfied. Then, some characteristic variables are selected from original variables through data preprocessing and descending dimension methods, which are defined as nodes that constitute the causal network. Second, the CCM-based methods are used to identify the causal direction and indirect causal relationship between variables, so as to construct the structure of the causal network. Since the CCM based on deterministic systems theory, it can handle nonlinearity and does not rely on the sample distribution. Finally, the weight of the edges in the graph is calculated to obtain the causal network which describes the process causality and serves as the basis for subsequent root causes tracing of alarms. The effectiveness of the proposed method is illustrated via a real industrial case study.
View