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.
Dear authors, we are pleased to verify that you meet the reviewer's valuable feedback to improve your research.
Thank you for considering PeerJ Computer Science and submitting your work.
Kind regards
PCoelho
[# PeerJ Staff Note - this decision was reviewed and approved by Mehmet Cunkas, a PeerJ Section Editor covering this Section #]
The authors correctly addressed all the reviewers' comments.
-
-
Comments are addressed.
-
-
**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.
**PeerJ Staff Note:** It is PeerJ policy that additional references suggested during the peer-review process should only be included if the authors agree that they are relevant and useful.
**Language Note:** When you prepare your next revision, please either (i) have a colleague who is proficient in English and familiar with the subject matter review your manuscript, or (ii) contact a professional editing service to review your manuscript. PeerJ can provide language editing services - you can contact us at [email protected] for pricing (be sure to provide your manuscript number and title). – PeerJ Staff
In this paper, the authors propose a novel deep learning framework for detecting friction-related failures in machinery using acoustic emission (AE) signals. The method integrates a Wasserstein Generative Adversarial Network with Gradient Penalty (WGAN-GP) for denoising noisy AE signals and a Long Short-Term Memory (LSTM) network for classifying fault types. The approach is structured in three main stages: signal enhancement, feature extraction via Short-Time Fourier Transform (STFT), and sequential classification. Experimental validation was conducted on three benchmark datasets (EAE-I, REB-II, TAD-III), achieving classification accuracies of up to 97%, significantly outperforming existing baseline models. The proposed model also shows strong generalization across different machine conditions and noise levels. An ablation study confirms that the GAN-based enhancement contributes a substantial performance boost of 7–9%. Performance metrics such as Precision (0.98), Recall (0.97), and F1-score (0.975) highlight the robustness of the system. The authors conclude that their GAN-LSTM framework offers an effective and scalable solution for real-time predictive maintenance and AE-based fault diagnosis, although they acknowledge challenges related to dataset diversity and computational demands for deployment on resource-limited devices.
The article exhibits a well-organized structure and clear presentation. The writing is clear, and the results are effectively communicated. The English is fine, even if it seems that the authors are not native speakers. A meticulous re-examination of the text would be beneficial to rectify any typographical errors and enhance the overall readability. The topic of the article is of both scientific and practical interest. The methodology is adequately described, and the results are compelling and well-presented. However, the article requires further revision before it can be considered for publication.
The reviewer’s comments are reported below:
- The in-text citations throughout the manuscript appear to be improperly formatted. In all instances, the publication year is replaced with "n.d." (no date), which suggests a systematic error, possibly due to incorrect use or export from the reference management software. This significantly affects the clarity and credibility of the literature references and should be corrected to ensure proper attribution and alignment with scholarly standards.
- In the Introduction, the authors state: “The limitation of the AE signal is that they are easily affected by noise and the environment with the signal processing techniques like the applications of the machine learning, deep learning, and generative models being necessary to improve the detections' accuracies and the classification of the faults (Kumar et al., n.d.)...”. However, this paragraph lacks a comprehensive discussion of the most established traditional (non-machine learning-based) signal processing techniques that have historically been used in the field of Acoustic Emission (AE), particularly for tasks such as signal detection and onset time estimation. For example, threshold-based methods, Akaike Information Criterion (AIC), and other classical statistical or heuristic approaches are widely adopted in AE analysis and should be acknowledged here. Including a concise overview of these foundational techniques would better contextualize the advancement represented by the proposed deep learning framework. To assist the authors, I suggest referring to the following publication as a starting point for structuring a more complete literature review of traditional AE signal processing methods. This reference provides a broad overview and cites various techniques and studies that the authors are encouraged to explore and cite appropriately:
o "Acoustic emission and artificial intelligence procedure for crack source localization." Sensors 23.2 (2023): 693.
- In the Introduction, the authors present a range of machine learning and deep learning techniques claimed to have been applied to acoustic emission (AE) signal analysis. While this overview is effective in drawing the reader's interest, it lacks the necessary support from a rigorous and comprehensive literature review. Many methods are mentioned—such as CNNs, LSTMs, GANs, and autoencoders — but proper citations are often missing, or references provided do not clearly demonstrate their application within the AE context. As it currently stands, the narrative is not sufficiently substantiated, leaving the reader without clear sources to consult for further understanding or verification. The reviewer recommends that the authors include appropriate and specific references to prior studies that have employed these methods in AE-related research. To support this improvement, a few example studies in which machine learning has been successfully applied to AE signal analysis are listed below. These should not be seen as exhaustive but can serve as a starting point, alongside the authors’ own sources, for building a more thorough and credible literature foundation:
o "Deep-Learning-Based Onset Time Precision in Acoustic Emission Non-Destructive Testing." 2024 IEEE International Workshop on Metrology for Living Environment (MetroLivEnv). IEEE, 2024.
o Identification of acoustic emission sources for structural health monitoring applications based on convolutional neural networks and deep transfer learning. Neurocomputing, 2021, 453: 1-12.
o Acoustic emission onset time detection for structural monitoring with U-Net neural network architecture. Developments in the Built Environment, 2024, 18: 100449.
**PeerJ Staff Note:** It is PeerJ policy that additional references suggested during the peer-review process should only be included if the authors are in agreement that they are relevant and useful.
- The manuscript contains several typographical and formatting issues that should be carefully addressed. For instance, there is a recurring omission of space between words and their corresponding in-text citation brackets throughout the document. Additionally, other minor typos and lexical inconsistencies are present in multiple sections. The authors are encouraged to perform a thorough proofreading of the manuscript to improve overall clarity, ensure grammatical accuracy, and adhere to academic writing conventions.
- The manuscript is structured with an excessive number of subsections, which may be more appropriate for a book chapter than for a scientific paper. This formatting choice interrupts the flow of the narrative and detracts from the readability of the article. The authors are encouraged to reduce the number of subsections by merging them where they are not strictly necessary. For example, Section 1.1 could be seamlessly integrated into the main Introduction without any loss of clarity, thereby improving the overall coherence and flow of the text.
- The reviewer observes that a substantial number of references included in the manuscript appear to be unrelated to the core subject matter of the paper, which focuses on the application of artificial intelligence to acoustic emission (AE) signal analysis for fault detection. Specifically, several cited works pertain instead to topics such as indoor air quality, sick building syndrome, HVAC systems, and biological or respiratory hazards, which do not appear to have any clear connection to AE signals, machine condition monitoring, or the proposed GAN-LSTM methodology. For example, the following references raise particular concern due to their apparent irrelevance:
o Huizenga, C., Abbaszadeh, S., Zagreus, L., & Arens, E. A. (n.d.). Air Quality and Thermal Comfort in Office Buildings: Results of a Large Indoor Environmental Quality Survey. Berkeley, CA, USA.
o Bluyssen, P. M., Cox, C., Seppänen, O., Oliveira Fernandes, E., Clausen, G., Müller, B., & Roulet, C. A. (n.d.). Why, when, and how do HVAC systems pollute the indoor environment, and what to do about it? The European AIRLESS Project. Build. Environ, 38, 209 225.
o Burge, P. S. (n.d.). Sick building syndrome. Occup. Environ. Med, 61, 185 190.
o Crook, B., & Burton, N. C. (n.d.). Indoor moulds, sick building syndrome and building related illness. Fungal Biol. Rev, 24, 106 113.
o Domínguez-Amarillo, S., Fernández-Agüera, J., Cesteros-García, S., & González-Lezcano, R. A. (n.d.). Bad air can also kill: Residential indoor air quality and pollutant exposure risk during the COVID-19 crisis. Int. J. Environ. Res. Public Health, 17, 7183.
o Dutkiewicz, J., Cisak, E., Sroka, J., Wójcik-Fatla, A., & Zajac, V. (n.d.). Biological agents as occupational hazards-selected issues. Ann. Agric. Environ. Med, 18, 286 293.
o Kreiss, K. (n.d.). The epidemiology of building-related complaints and illness. Occup. Med, 4, 575 592.
o Kumar, P., Martani, C., Morawska, L., Norford, L., Choudhary, R., Bell, M., & Leach, M. (n.d.). Indoor air quality and energy management through real-time sensing in commercial buildings. Energy Build, 111, 145 153.
o Lei, L., Chen, W., Xue, Y., & Liu, W. (n.d.). A comprehensive evaluation method for indoor air quality of buildings based on rough sets and a wavelet neural network. Build. Environ, 162, 106296.
o Mannan, M., & Al-Ghamdi, S. G. (n.d.). Indoor air quality in buildings: A comprehensive review on the factors influencing air pollution in residential and commercial structure. Int. J. Environ. Res. Public Health, 18, 3276.
o Pérez-Padilla, R., Schilmann, A., & Riojas-Rodriguez, H. (n.d.). Respiratory health effects of indoor air pollution. Int. J. Tuberc. Lung Dis, 14, 10791086.
o Pietrogrande, M. C., Casari, L., Demaria, G., & Russo, M. (n.d.). Indoor air quality in domestic environments during periods close to Italian COVID-19 lockdown. Int. J. Environ. Res. Public Health, 18, 4060.
o Santos, J., Ramos, C., Vaz-Velho, M., & Vasconcelos Pinto, M. (n.d.). Occupational exposure to biological agents. International Conference on Applied Human Factors and Ergonomics, 6167.
o Urrutia-Pereira, M., Mello-da-Silva, C. A., & Solé, D. (n.d.). Household pollution and COVID-19: Irrelevant association? Allergol. Et Immunopathol, 49, 146149.
o Wyon, D. P. (n.d.). The effects of indoor air quality on performance and productivity. Indoor Air, 14, 92 101.
It is unclear whether their inclusion is the result of an honest error—perhaps due to a mismanaged reference manager—or whether this reflects an attempt to artificially inflate citation counts or distract from a lack of relevant literature review. Regardless of the reason, these references do not support the claims made in the manuscript and do not pertain to the field of AE signal processing or AI-based fault detection. The reviewer strongly recommends that the authors either (1) justify and clarify the relevance of each of these references within the context of the current study, or (2) remove them from the manuscript entirely. Additionally, the authors should revise the literature review to ensure that it draws upon genuinely relevant and authoritative sources in the fields of acoustic emission, machine learning, signal processing, and predictive maintenance.
- Throughout the manuscript, the mathematical expressions and equations are not always accompanied by adequate definitions of the variables and terms used. In several instances, key symbols introduced in equations are not explicitly defined in the surrounding text, which compromises clarity and hinders the reader's ability to fully understand the formulation. This issue appears consistently across the manuscript. The authors are strongly encouraged to perform a detailed check of all equations and ensure that every variable, symbol, and parameter is clearly introduced and defined immediately before or after it appears in the equation.
- Most of the figures in the manuscript appear to be of low visual quality, with noticeable pixelation and poor resolution, which detracts from the clarity and professional presentation of the work. The authors are strongly encouraged to increase the resolution of all figures, or preferably, to replace them with vector-based graphics (e.g., in PDF, EPS, or SVG format) to ensure that all elements—particularly text, labels, and data curves—remain sharp and legible in both print and digital formats. This improvement is essential for effective visual communication of the results and to meet publication standards.
Methods described with sufficient detail & information to replicate, but the sources are not adequately cited.
-
In conclusion, the paper presents a highly interesting and relevant study, particularly in terms of its methodological approach and potential applications. However, the current presentation of the results is inadequate and significantly undermines the clarity and impact of the work. Therefore, the reviewer recommends the manuscript for publication only after major revisions, as outlined above.
1) The results listed in Fig.7 show a relatively limited improvement in performance compared to previous studies. Thus, the engineering application potential needs to be further explored.
2) The listed references are out-of-date (most before 2020), and transfer learning and ensemble learning are the hot topics in the AI-based academic field, but the relevant works are rarely analyzed.
3) All basic knowledge should be introduced via a new section; the methodology part only should introduce the original innovation content of the proposed method, which is beneficial to clearly show the contributions of this paper.
4) The limitations and further works of this paper should be added.
5) More comparative methods should be added to further verify the superiority of the proposed method.
-
-
This paper proposes a deep learning-based GAN-LSTM framework for friction failure detection and acoustic emission signal enhancement in machine engines. In general, the paper is well presented. However, the following issues must be addressed to improve its clarity, scientific rigor, and relevance:
1. More background and motivation should be added to the introduction to ensure accessibility for readers who may not be familiar with the topic.
2. Descriptions of well-known concepts can be reduced to avoid redundancy and maintain focus on the novel contributions of the work.
3. The rationale for introducing the failure detection method must be clearly explained. The authors should highlight the major advantages of their approach compared to traditional methods.
4. Some related work on this topic may be read, for example “Federated Transfer Learning for Remaining Useful Life Prediction in Prognostics with Data Privacy,” “Data-driven deep learning approach for thrust prediction of solid rocket motors,” and “Neuromorphic Computing-Enabled Generalized Machine Fault Diagnosis With Dynamic Vision” should be considered to strengthen the literature foundation.
**PeerJ Staff Note:** It is PeerJ policy that additional references suggested during the peer-review process should only be included if the authors agree that they are relevant and useful.
5. Ablation studies must be included to evaluate the impact of key parameters in the proposed method. This is essential to validate the robustness and effectiveness of the framework.
-
-
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.