Detection and Location of Model-Plant Mismatch in Multiple Input Multiple Output Systems under Model Predictive Controller Using Granger Causality Method

Round 1

Reviewer 1 Report

The paper is a valuable contribution to the detection and location of plant-model mismatsch in MPC controlled multivariable systems. I have only few recommendations for a possible improvement of the text.

1) English language and style must be improved significantly, since some mistakes seriously worsen the readability and can even lead to confusion of the reader.

2) Since the Granger causality method is compared with decussation (CAID and MQI) which are familiar only to specialists in the control performance monitoring community, it seems not sufficient to only cite the source (Li et al.). Instead, calculation of CAID and MQI should shortly be repeated in your paper.

3) In addition to the Wood-Berry column example, a more realistic MPC control problem should be treated to demonstarte the applicability of the newly developed model under industrial conditions. It would also lead to a better impression about how much the complexity of calculations will be increased when applying the method to more complex processes.
The example selected should include
a) more manipulated and control variables as well as measured disturbances,
b) different models for sub-processes such as higher order, integrating and zero gain models (no influenece of ui on yj)
c) nonlinear GP model, linear GM model.

4) A few remarks should be given regarding the sensitivity of the method to changes in the MPC contoller tuning parameters.

Author Response

Please see the attachment.

Reviewer 2 Report

Following are my observations.

This paper proposed a Model Plant Mismatch (MPM) detection method for closed-loop Multiple Input and Multiple Output (MIMO) process control systems based on Granger Causality method.

Authors had tried to address an already identified issue in process control systems and took necessary steps to demonstrate the proof of the method. In this perspective I really appreciate their attempt. However, when presenting their work they need to follow the basic guidelines as well as journal guidelines.  Unfortunately, when presenting this paper authors were unable to convince me on those perspectives. Throughout the paper it was observed inconsistency and sometimes carelessness. Most importantly, connection between sections in the paper is very poor. Following I pointed out several such points.

Title

1. The title of this paper is “Detection and location of model-plant mismatch in MIMO closed-loop systems under MPC controller using Granger causality method”. In the title abbreviations are directly used (MIMO, PMC) and this will make difficulties to non-subject experts to understand the context of the paper. Therefore, I proposed to do the required changes in the title.

Abstract

1. Also in the abstract undefined abbreviations (MIMO, MPC, CAID, MQI) are used. This must be corrected.
2. There is a mismatch in the title and the content of the abstract. In title authors mentioned about “MPC” and there is no any information about MPC in the abstract. Please do necessary changes in the abstract.

Introduction

Line 20: “process of production in chemistry and industry”. - Idea is not clear, rewrite the sentence.

Line 21: Abbreviation “PID” is used without defining in the first time and not even visible in abbreviation list. Check all similar situations and update accordingly.

Line 25: Thus far, fault diagnosis has gradually come to be a hot topic in process monitoring, diagnosis, analysis and management in recent decades. – Inappropriate statement, because fault diagnosis is always critical in any process monitoring system.

Line 26: It is challengeable that most diagnosis technologies rely more on data driven methodology rather than model based framework so …. – Distinguish between “data driven” and “model based” with references. However, there is a under laying data in any model driven system. Thus, use appropriate terms and update the sentence.

Line 33 and Line 45: Minimum Variance benchmark (MV) is present. It is acceptable to have abbreviations with the definition in different sections for improve the readability (in line 35 only MV is present). However, here, it appears in the same paragraph in inconsistent manner. Address this and similar situations.

 

Please read the paper again carefully and you can find several equal situations throughout the paper and correct and update.

 

Illustrations

1. In all the figures the caption is a non-descriptive caption. However, caption must be strong enough to convey the idea represents by the illustration (descriptive and comprehensive caption).
2. Some figures are (e.g.: Figure 5, Figure 9) without legends.

 

Discussion

Poorly written discussion. My suggestion is to merge the result and discussion.

Conclusion

Considerable portion of this section is allocated to mention the advantages of “Granger causality method” but not about the method tried to introduced in this paper.

Abbreviations List

1. Authors include a section for Abbreviations (in line 567). However, there are many missing abbreviations, and the list must be updated.

References

1. References in the list are not in the format required by the journal: in most of the references et al. is used. Update the references according to the format of the journal.

   

 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I still believe that a more realistic (not 2MVx2CV) MPC problem should be included in the paper to demonstrate the applicability of the methods described under industrial conditions. More realistic means
- more the 2 MVs and CVs
- both setpoint and zone control
- more complex and different models in the MV-CV channels
  such as higher order, integrating, and zero gain models
- nonlinear GP, linear GM model

Author Response

See the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

In “Response to the Reviewer Comments” authors stated that they updated the manuscript as follows.  However, in the manuscript it is not visible. Therefore, please check the all the stated updates are visible in the actual manuscript.

Line 26: It is challengeable that most diagnosis technologies rely more on data driven methodology rather than model based framework so …. – Distinguish between “data driven” and “model based” with references. However, there is a under laying data in any model driven system. Thus, use appropriate terms and update the sentence.

Thank you for this valuable comment. Model-based diagnosis methods rely on a model that defines nominal behavior of a dynamic system to detect abnormal behaviors and isolate faults. On the other hand, data-driven diagnosis algorithms detect and isolate system faults by operating exclusively on system measurements and using very little knowledge about the system. We have added the following reference the manuscript.

Khorasgani, H., Farahat, A., Ristovski, K., Gupta, C., & Biswas, G. (2018). A Framework for Unifying Model-based and Data-driven Fault Diagnosis. Annual Conference of the PHM Society, 10(1)

 

Comments for author File: Comments.pdf

Author Response

see the attachment

Author Response File: Author Response.pdf