Overview of Machine Learning Methods for Lithium-Ion Battery Remaining Useful Lifetime Prediction

Round 1

Reviewer 1 Report

This paper renders some ML-empowered techniques to predict lithium-Ion battery remaining useful lifetime. Regarding the significance of the role of li-ion batteries in EVs and smart grids, the idea of the paper is timely. Nevertheless, as a review paper, there are some major issues that need to be addressed as follows:

1- Authors need to prepare a comprehensive flowchart to describe the methods used in the understudied papers considering all important parts of the ML methods. 

2- Some information about the whole number of papers in this field and the way of reifying them need to be added to the manuscript. This way, authors should explain how they select the final papers.

3-  It is essential to summarize and conclude the explanations of each subsection in section 2 in order not to confuse the readers. For example, 2.1. Support Vector Machine / 2.2. Gaussian Process Regression / 2.3. Extreme Learning Machine / 2.4. Deep Neural Network 299

4- For a review paper, the literature review is very poor. The authors need to use some important publications in the aspect of applications of ML li-ion batteries as follows:
a. An intelligent approach for nonlinear system identification of a Li-ion battery  DOI: 10.1109/I2CACIS.2017.8239040
b. A dynamic artificial neural network approach to estimate thermal behaviors of li-ion batteries DOI: 10.1109/I2CACIS.2017.8239043
c. Using a soft computing method for impedance modeling of Li-ion battery current DOI: 10.1504/IJAIP.2020.106686
d. An ANFIS Approach to Modeling a Small Satellite Power Source of NASA DOI: 10.1109/ICNSC.2019.8743333

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

The paper is a review of Machine Learning Methods for Lithium-Ion Batteries.

1) There have been a few other recent reviews covering the same topic of Machine Learning Methods for Lithium-Ion Batteries (i.e. https://doi.org/10.3390/electronics10111309)
Which is the additional value given by your manuscript?

2)For this purpose, I suggest citing and list in the reference list  more (recent) papers on the same topic
Notice that no limit of pages in MDPI publication.
For instance, the recent paper from MDPI, with overviews on the BMS such as Technology (i.e. https://doi.org/10.3390/technologies9020028), Energy (https://doi.org/10.3390/en4111840), Sensor (https://doi.org/10.3390/s21175698) and Applied Science ( https://doi.org/10.3390/app8040534) should be cited. Same for
Electronics (https://doi.org/10.3390/electronics10151859  https://doi.org/10.3390/electronics10111309 https://doi.org/10.3390/electronics9030510  
and other special issue paper https://www.mdpi.com/journal/electronics/special_issues/Battery_Management_System) where the importance of the precise charge in any condition is shown, or also other from IEEE (i.e. 10.1109/EMCEurope.2019.8872061)
Please carefully look at the literature and increase the reference list (even doubling it would be appreciated).

3) Please stress more along with the paper about real-time monitoring together with prediction being this a safety-critical topic. For instance, the importance of charge equalizing in any operating condition should be stressed and how a prevision method could be useful instead of deterministic and real-time monitoring. To the best of my knowledge, a precise charge in terms of less than tens of mV should be guaranteed otherwise the battery pack could also explode and lifetime prediction could be interrelated with this issue. 
I believe that constant monitoring is the one that better faces the safety constraints. Please comment on this

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

In this research work, the authors, classify, and compare different machine learning (ML) based 18 methods for the prediction of the RUL of Li-ion batteries. First, this article summarizes and classifies 19 various methods proposed in recent years to predict the RUL of Li-ion batteries using ML. Although, the results look promising, but there are some comments that should be addressed before publishing in Materials.

 

 

• I commend the authors to add some more information about the results in abstract and its better to short the introduction in the start of the abstract.

 

• The English and some minor typo errors could be improved.

Remarks: Published after solving all the above minor comments/concerns.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

All comments have been addressed, and the paper can be considered for publication

Reviewer 2 Report

The paper is suitable for publication

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

As the authors noted, there have been a few other recent reviews covering similar topics. The present manuscript fails to demonstrate necessity of an additional literature survey. It is neither clear what new conclusions can be made from the present review work. How does this work further direct future research and application? The manuscript demands significant re-writing to clearly illustrate the purpose and significance of this latest reviewing effort. 

Reviewer 2 Report

The paper is a review of Machine Learning Methods for Lithium-Ion Batteries.

 

1) The authors refer to what is usually named as "Battery Management system" (BMS) even if this keyword is totally not considered along with the paper. Please explicitly mention this keyword so that your paper could be easily reached by potential readers

2)The described model review is interesting. Anyhow, I would suggest stressing more along with the paper about real-time monitoring together with prediction being this a safety-critical topic. For this reason, I suggest below some papers should be included in the paper.

3)The paper is interesting and being a review, I like the presence of any references. Nevertheless, I would imagine to find other recent ones on the same topic (i..e
https://doi.org/10.3390/electronics10070846).
Notice that no limit of pages in MDPI publication.
For instance, the recent paper from MDPI, with overviews on the BMS such as Technology (i.e. https://doi.org/10.3390/technologies9020028), Energy (https://doi.org/10.3390/en4111840), and Applied Science ( https://doi.org/10.3390/app8040534) should be cited. Same for
Electronics ( https://doi.org/10.3390/electronics10030293 ,  https://doi.org/10.3390/electronics9030510, https://doi.org/10.3390/electronics10060705
https://doi.org/10.3390/electronics10030293
and other special issue paper https://www.mdpi.com/journal/electronics/special_issues/Battery_Management_System) where the importance of the precise charge in any condition is shown, or also other from IEEE (i.e. 10.1109/EMCEurope.2019.8872061 ,  10.1109/ISEMC.2015.7256257)

4)The suggested above-mentioned papers also highlight the safety tasks that are critical and not properly considered along with the paper. The importance of charge equalizing in any operating condition should be stressed and how a prevision method could be useful instead of deterministic and real-time monitoring. To the best of my knowledge, a precise charge in terms of less than 10mV should be guarantee otherwise the battery pack could also explode and lifetime prediction could be interrelated with this issue. 
I believe that constant monitoring is the one that better faces the safety constraints. Please comment on this.