Next Article in Journal
Planning and Optimizing Charging Infrastructure and Scheduling in Smart Grids with PyPSA-LOPF: A Case Study at Cadi Ayyad University
Previous Article in Journal
A Road-Adaptive Vibration Reduction System with Fuzzy PI Control Approach for Electric Bicycles
Previous Article in Special Issue
Experimental Study on Distributed Measurement of Internal Pressure in Lithium-Ion Batteries Using Thin-Film Sensors
 
 
Article
Peer-Review Record

Improving State-of-Health Estimation for Lithium-Ion Batteries Based on a Generative Adversarial Network and Partial Discharge Profiles

World Electr. Veh. J. 2025, 16(5), 277; https://doi.org/10.3390/wevj16050277
by Hangyu Zhang and Yi-Horng Lai *
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3:
World Electr. Veh. J. 2025, 16(5), 277; https://doi.org/10.3390/wevj16050277
Submission received: 21 February 2025 / Revised: 29 April 2025 / Accepted: 14 May 2025 / Published: 16 May 2025
(This article belongs to the Special Issue Lithium-Ion Battery Diagnosis: Health and Safety)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Title: Improving State of Health Estimation for Lithium-ion Batteries based on GAN and Partial Discharge Profiles:     

The manuscript entitled " presents a method for lithium-ion battery SOH (State of Health) estimation by combining DoppelGANger and Temporal Convolutional Networks (TCN)"   

Overall, the manuscript will need to be significantly changed to meet the journal's requirements while also grabbing the attention of the readers. Some of the issues that must be resolved prior to publication are mentioned below:   

  1. Grammar polishing and proof reading should be done for the whole content of the manuscript.
  2. Minor modifications and formatting should be done for Figures. (e.g. Fig. 2, needs better axis labeling for SoC ranges, Fig. 1, lacks better readability and proper axis labeling).
  3. Please incorporate additional citations. Resources used are very limited.
  4. The Data augmentation performed is not justified. How does the GAN-generated dataset compare in distribution similarity metrics (e.g., Wasserstein distance, KL divergence) with the real dataset?
  5. How would your proposed model perform in real-world EVs where battery degradation can be influenced by temperature, driving habits, and charging cycles?
  6. How does the proposed approach compare with physics-informed machine learning models? Would be interesting if you incorporate sort of comparisons to make your result conclusions more viable.
  7. Also, more quantitative evaluations would be required for synthetic data validation.
Comments on the Quality of English Language

English Language needs refinement for better Clarity 

Author Response

We gratefully thank the editor and all reviewers for their time spent making their constructive remarks and useful suggestions, which have significantly raised the quality of the manuscript and have enabled us to improve the manuscript. Each suggested revision and comment brought forward by the reviewers was accurately incorporated and considered. Please refer to the attached file. 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This work describes an NN developed to estimate the SOH of batteries from partial discharge data. The aim is interesting and well-described. The first 5 sections are well written, but more citations are needed; there is some general description of the network or the procedures that must be reinforced by a reference. Another reference that must be presented is the link to the NASA dataset.

In these sections, very few minor suggestions can be made. My more relevant ones are:

I suggest the authors replace the x-axis "capacity" label with a more appropriate label, "Moved Charge." The data represent a charge moved, not the battery's capacity or remaining capacity. 

Formula 6 contains a typo. The authors multiply a difference by 100%, which is not correct; the normalization of the difference is probably not reported in the formula.

Figure 7 is not a demonstration; this figure shows the procedure.

The main problem with the work is the last two sections. The authors quickly report the major indicators relative to the results and make a few comments about that. These sections completely lack a comparison with the literature. The results alone are insufficient to evaluate whether the presented methodology is acceptable. The 10% error reach can be unacceptable if a better estimation method exists. Moreover, the authors claim that their method is better than other methodologies with a low SOC, but they do not provide results or a comparison to prove it. Finally, a comment to describe because, in the first 2 sets, the raw data work better than the synthetic ones should be provided.

 

 

Author Response

We gratefully thank the editor and all reviewers for their time spent making their constructive remarks and useful suggestions, which have significantly raised the quality of the manuscript and have enabled us to improve the manuscript. Each suggested revision and comment brought forward by the reviewers was accurately incorporated and considered. Please refer to the attached file.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This paper proposes a State of Health (SOH) estimation model that combines DoppelGANger and Temporal Convolutional Networks (TCN).
However, the manuscript requires significant improvement before it can be considered for publication in its current form.

- All abbreviations should be defined at their first mention and used consistently throughout the manuscript.
- The literature review requires expansion to provide a more comprehensive reference analysis.
- The novelty and main contributions of the proposed model are not clearly stated and appear to overlap with those of reference [14]. This point needs clarification, along with a well-structured comparison highlighting the differences and improvements.
- To better highlight the contributions, some sentences should be added to the introduction section, preferably in an itemized format.
- A constructive research gap should be clearly presented just before the main contributions.
- Simulation results should include comparison with the basic TCN or work in reference [14].
- The quality of the figure 12 is low. Also, it is recommended to enhance its resolution and consider plotting both the raw and synthetic data in a single figure for clearer visual comparison.
- The conclusion shall be more constructive. Limitation of current approach and future solution need to be included.

Author Response

We gratefully thank the editor and all reviewers for their time spent making their constructive remarks and useful suggestions, which have significantly raised the quality of the manuscript and have enabled us to improve the manuscript. Each suggested revision and comment brought forward by the reviewers was accurately incorporated and considered. Please refer to the attached file.

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The new version provides an interesting step forward in the SOH estimation using NN. The paper is well-written, and the references are adequate. Now is a complete and interesting work that should be published.

Reviewer 3 Report

Comments and Suggestions for Authors

I appreciate the authors' efforts in revising the manuscript. They have carefully considered my comments and provided satisfactory responses to my concerns.

Back to TopTop