All reviews of published articles are made public. This includes manuscript files, peer review comments, author rebuttals and revised materials. Note: This was optional for articles submitted before 13 February 2023.
Peer reviewers are encouraged (but not required) to provide their names to the authors when submitting their peer review. If they agree to provide their name, then their personal profile page will reflect a public acknowledgment that they performed a review (even if the article is rejected). If the article is accepted, then reviewers who provided their name will be associated with the article itself.
The paper may be accepted.
[# PeerJ Staff Note - this decision was reviewed and approved by Maurice H. ter Beek, a PeerJ Section Editor covering this Section #]
All the comments are addressed satisfactorily, and no more comments from my side.
N/A
N/A
N/A
Incorporate the comments of the reviewers.
**Language Note:** The review process has identified that the English language must be improved. PeerJ can provide language editing services - please contact us at [email protected] for pricing (be sure to provide your manuscript number and title). Alternatively, you should make your own arrangements to improve the language quality and provide details in your response letter. – PeerJ Staff
All issues are addressed well.
nothing
nothing
nothing
The authors addressed most of the comments satisfactorily. However, a few more minor comments need to be addressed before considering it final from my side.
No further comments. Already addressed.
No further comments. Already addressed.
1) The quality of English and figures needs to be improved in order to meet the standards of the journal. For instance, although the English is generally clear, a few sentences—especially in the Introduction (lines 20–40) and Conclusions (lines 460–470)—contain long or awkward phrasing. A brief grammatical polish by a native or professional editor would enhance readability.
2) Some figures (e.g., Figs. 5–8) have small axis labels and legends that are difficult to read. Please enlarge font sizes and ensure consistent units across figures.
3) It is recommended to discuss recent works like Advanced Fault Diagnosis in Milling Machines Using Acoustic Emission and Transfer Learning and A Hybrid Deep Learning Framework for Fault Diagnosis in Milling Machines, etc., in literature.
4) Literature needs a little refinement while adding references of recent papers from 2024 and 2025 mostly to meet the trends of the current time.
It is recommended to extend the literature further by discussing different approaches used for feature extraction, fusion, and multimodal.
The revised version of the manuscript has addressed almost all the comments. A few more are added to enhance the quality of the manuscript.
-
1) The main contribution and the reasons (discussed) just before that, the introduction section needs to be elaborated more.
2) Discuss the study limitations and future recommendations.
3) The quality of English needs to be improved, and a thorough review is recommended to remove minor grammatical mistakes.
It is recommended to extend the literature further by discussing different approaches used for feature extraction, fusion, and multimodal.
**PeerJ Staff Note:** Please ensure that all review, editorial, and staff comments are addressed in a response letter and that any edits or clarifications mentioned in the letter are also inserted into the revised manuscript where appropriate.
The authors have satisfactorily addressed the previous comments.
Addressed
Addressed
1. Explicitly state the train/validation/test split protocol (e.g., folds, subject/machine separation, windowing strategy), random seed(s), and number of repeated runs used for the reported metrics.
2. For the main results table/figure, add variability (mean ± SD or 95% CI) across folds/runs and harmonize significant figures between text and tables.
3. It is recommended to discuss A Hybrid Deep Learning Approach for Bearing Fault Diagnosis Using Continuous Wavelet Transform and Attention-Enhanced Spatiotemporal Feature Extraction and Advanced Fault Diagnosis in Milling Machines Using Acoustic Emission and Transfer Learning in the introduction.
4. Upload higher-resolution figures (axes/legends legible at 100%) and ensure consistent units and parameters (Hz vs kHz, ms vs s, sampling rate, window length/overlap). Include a brief data/code availability statement.
**PeerJ Staff Note:** Please ensure that all review and editorial comments are addressed in a response letter and that any edits or clarifications mentioned in the letter are also inserted into the revised manuscript where appropriate.
**Language Note:** The review process has identified that the English language must be improved. PeerJ can provide language editing services - please contact us at [email protected] for pricing (be sure to provide your manuscript number and title). Alternatively, you should make your own arrangements to improve the language quality and provide details in your response letter. – PeerJ Staff
-
-
-
I have read your paper carefully, and some comments are listed as follows:
1. The introduction section is too long for a technical paper. It is recommended to follow a four-paragraph approach as follows: 1) State the motivation and describe the problem to solve. 2) Revise the state-of-the-art with reference to published key works. 3) Introduce your proposal in a fair context to other published methodologies. 4) Clearly state the novelty/contribution of the proposed methodology and briefly describe how it is validated in context to other methodologies. All technical details and figures must be placed elsewhere.
2. The literature review is incomplete and out-of-date (all before 2020). I suggest the authors rewrite the review according to the development timeline of fault diagnosis technology (traditional machine learning → deep learning → transfer learning). Some latest works should be cited, which can be searched by Google Scholar.
3. Try to clearly state the contribution of your proposal in a fair context to other published methodologies (i.e., make a comparison between your proposal and other methodologies to see the advantages)
4. All basic knowledge should be introduced via a new section; the methodology part should only introduce the original innovation content of the proposed method, which is beneficial to clearly show the contributions of this paper.
The manuscript presents a novel expert system for compressor fault diagnosis that integrates PCA-SPE-CNN-based deep learning with a dynamic knowledge graph and natural language processing (NLP) inference. The proposed method aims to enhance real-time abnormality detection, reduce false alarms, and enable effective fault tracing using petrochemical instrumentation data. Experimental validation on real compressor data shows high diagnostic accuracy and effective source localization of faults.
• The use of a sliding window-based fault-tolerant filter combined with PCA-SPE is innovative; however, more clarity is needed on how hyperparameters (e.g., window size, filter residual thresholds) were optimized.
• The CNN structure and parameter tuning are well-described, but the rationale for choosing small kernels (1×3) over larger ones and the impact on model performance should be justified quantitatively.
• The model achieves perfect classification performance (100% accuracy, AUC = 1), which raises concerns about overfitting. Are external datasets or unseen test conditions used to validate generalization?
• Although comparisons are provided with KNN, SVM, and CNN models, other state-of-the-art deep learning approaches (e.g., LSTM, Transformer-based models) should be included for a comprehensive benchmark.
• The paper claims the system is suitable for real-time monitoring, but no latency or computational load analysis is provided. Please include inference time per sample and system hardware specifications.
• The dynamic knowledge graph construction is impressive, but its incremental learning capabilities and integration with streaming data need further elaboration. How is outdated or conflicting knowledge resolved?
• The method’s applicability beyond compressors (e.g., to other rotating machinery or industries) is mentioned briefly. A more in-depth discussion of limitations and adaptability to different domains would strengthen the conclusion.
• It is recommended to discuss possible limitations such as reliance on labeled fault data, assumptions of Gaussian distributions in PCA, or the system's performance under sensor noise or data loss. Future directions may include self-supervised learning or federated monitoring frameworks.
• It is recommended to discuss recent papers in the introduction section of the manuscript.
All text and materials provided via this peer-review history page are made available under a Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.