A Reduced Order Model of the Thermal Profile of the Rolls for the Real-Time Control System

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper „A Reduced Order Model of the Thermal Profile of the Rolls for the Real-Time Control System“ analyzes the approaches to modeling thermal profile of rolling-mill work rolls: finite-element model, the results of which are used for training the developed thermal model based on the neural network and a reduced-order model based on fitting low-order transfer functions to the FE results. The use is demonstrated over a full rolling campaign and parameter dependencies are discussed.

Overall, the paper is relevant, well-written, methodology is clear and comparisons are comprehensive. However, certain aspects need to be addressed before the paper can be accepted.

1. Is there any possibility that experimental validation could be performed?
2. What are the limitations of the model and the outlook for future work?
3. The notation is inconsistent in places (e.g. kn and k_n). Please edit
4. If the neural network metamodel is ultimately discarded in favor of the ROM, it would be advisable to compress its description.

Author Response

Replay is in the attached file.

Author Response File: Author Response.docx

Reviewer 2 Report

Comments and Suggestions for Authors

The author presented the thermal profile models of the rolls for a real-time control system for the design of the rolling schedule.

Below are comments for improving the manuscript:

1. The abstract should be complemented with quantitative results, which would better highlight the conducted research.

2. Simultaneous closed referencing of multiple sources is not recommended (line 36).

3. Since the field is rapidly evolving and renewing, it would be relevant to supplement the introduction with the most recent research from the last five years.

4. Although the author states the research objective, it is also important to describe the scientific novelty and practical relevance in detail. The literature review should formulate the scientific gap.

5. Neural network (NN) models are mentioned in the review — how are they connected to your work? This should be described in more detail in the Introduction section.

6. What is the reference for equation (6)?

7. The abbreviation MMNN should be explained in the text.

8. How do your results correlate with those of other researchers?

9. The conclusions should be improved to reveal the main results more clearly. They should also be supplemented with future research plans and necessary studies in this field.

Author Response

Replay is in the attached file.

Author Response File: Author Response.docx

Reviewer 3 Report

Comments and Suggestions for Authors

This study focuses on developing fast and accurate models for the thermal profile of rolls in flat rolling processes, crucial for real-time control systems. The study proposes semi-analytical, Finite Element (FE), Neural Network (NN)-based metamodel (MMNN), and Reduced Order Model (ROM) based on transfer functions to address the computational demands of online control. This study is pretty interesting; however, some major suggestions are provided to the authors to improve it.

Major suggestions

 

1. Suggested to add more quantitative data in the abstract.
2. To strengthen the introduction section, consider incorporating recent publications (especially after 2020) to provide a critical review of the latest and related works. This approach will demonstrate a comprehensive understanding of current advancements in the field and highlight the paper's relevance to contemporary research discussions. You may consider adding more literature (Input Attribute Optimization for Thermal Deformation of Machine-Tool Spindles using Artificial Intelligence; Coolant Volume Prediction for Spindle Cooler with Adaptive Neuro-fuzzy Inference System Control Method; Process Parameter Optimization in Czochralski growth of silicon ingots: a Monte Carlo-finite element coupled model; Data-driven approach for optimizing the Czochralski process and predictive modeling: a finite element and machine learning analysis).
3. Suggested to add a paragraph in the introduction highlighting the research gap and their needs, and then outline the novelty and its practical implications for industrial control systems.
4. Suggested to add a detailed paragraph to elaborate the choosing the transfer function Rom over the MMNN for the thermal profile calculation.
5. Suggested to add a table for the data collected for the modeling and how much data is used, the data splitting percentage for training and testing, algorithms-related parameter settings and the stopping criteria (for your reference, you can go through the article: Input Attribute Optimization for Thermal Deformation of Machine-Tool Spindles using Artificial Intelligence).
6. The effect of changing the control parameters of each algorithm on the performance of the problem should be investigated and reported.
7. Suggested to add a flowchart of the proposed model for a better presentation and understanding of readers.
8. Suggested to add explanation of the identification method used to extract these parameters from the FEM simulation data.
9. Suggetsed to add a detailed discussion of coupling of thermal and thermal expension FEM model and how the temperature distribution from the chosen ROM (transfer function) is integrated into the thermal expansion FEM model (Equation 29). This is a critical aspect for understanding the complete real-time control system and requires further explanation.
10. Suggested to add a table for comparative performance of each model (MMNN vs. ROM vs. FEM) and justify the selection of the ROM for practical use.
11. Suggested to add a detailed explanation of how the proposed ROM improves rolling schedule optimization, enhances product flatness, reduces energy consumption, or minimizes material waste compared to current practices.
12. Suggested to add a discussion on potential implementation pathways in industrial settings, including data acquisition, latency considerations, and control system integration.

Author Response

Replay is in the attached file.

Author Response File: Author Response.docx

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

The revised manuscript seems ok. No further suggestions.