Fast Health State Estimation of Lead–Acid Batteries Based on Multi-Time Constant Current Charging Curve

Round 1

Reviewer 1 Report

The paper presents the application of long short-term memory regression model optimised by a bat algorithm to estimate the SoH of lead-acid batteries. 

From the paper, it is not clear how the proposed methodology directly relates to a battery's properties. Even though the battery is charged and discharged in the laboratory environment with a constant discharge current the presented model fails to precisely follow the real capacity as shown in Fig. 5. 

The paper sounds as if the authors tried to find the application for the described algorithm rather than find a way to improve the state of art in SoH prediction for lead-acid batteries.

Moderate editing of English language required.

Author Response

Dear reviewer:

Thank you for your patient review and valuable suggestions. Here is our response to your Comments and Suggestions:

Comment/Suggestion 1: The paper presents the application of long short-term memory regression model optimised by a bat algorithm to estimate the SoH of lead-acid batteries. 

Response:

At present, some literatures have regarded battery health status prediction as time series prediction, and also proved that using swarm intelligence algorithm combined with machine learning algorithm to predict time series has better prediction performance, such as literature [13]. Therefore, this paper uses bat algorithm combined with LSTM network to predict the health state of battery, and the experimental results also proved the feasibility of the proposed algorithm.

Comment/Suggestion 2: From the paper, it is not clear how the proposed methodology directly relates to a battery's properties. Even though the battery is charged and discharged in the laboratory environment with a constant discharge current the presented model fails to precisely follow the real capacity as shown in Fig. 5. 

Response:

Since the attenuation change of lead-acid battery is more complex than that of lithium battery, and the attenuation curve is more irregular, the prediction effect is slightly weaker than that of lithium battery.

Comment/Suggestion 3: The paper sounds as if the authors tried to find the application for the described algorithm rather than find a way to improve the state of art in SoH prediction for lead-acid batteries.

Response:

Thank you for your patient review and valuable suggestions. 

Due to the limited literature on the predictive performance of lead-acid batteries and we couldn’t find a suitable publicly available experiment data about to lead-acid batteries, this article lacks comparative analysis with other prediction methods.

 In order to improve this situation, we have added some experimental analysis content in the article, and the predicted performance obtained from the experimental results is still acceptable in practical applications.

Author Response File: Author Response.pdf

Reviewer 2 Report

This manuscript proposes a model that couples the long short-term memory (LSTM) and the Bat Algorithm (BA) to estimate the state of health (SOH) of batteries through charging and discharging cycles. The methodology is well presented. However, a few concerns may need to be addressed before it is accepted.

1. Only one set of results is presented (Fig. 5), where the improvement of the prediction from the proposed model seems to be quite limited. The main feature (a drop in the measured data) is not captured. In addition, the explanation of the result should be enriched.

2. The prediction performance is not superior compared to other’s work, i.e., https://doi.org/10.1016/j.energy.2023.127585, which may let the reader doubts the novelty of this work.

3. The SOH estimation is usually implemented up to thousands of cycles, or below 80% of the original capacity. However, it only 50 cycles with 88.1% (26/29.5) remanning. 

4. The measured degradation curve looks strange. Usually, it is a nonlinear one with an initial slowing decreasing range followed by a rapid fading one after the knee point.  It is more convincing to use publicly available experiment data, i.e., the NASA dataset (Saha B, Goebel K. Battery data set. NASA AMES prognostics data repository 2007.), Oxford dataset (Birkl C. Diagnosis and prognosis of degradation in lithium-ion batteries. University of Oxford; 2017), etc.

5. There is not any work cited for the bat algorithm.

The writing can be improved.

Author Response

Dear reviewer:

Thank you for your patient review and valuable suggestions. We have tryed to improve the English writing. Here is our response to your Comments and Suggestions:

Comment/Suggestion 1: Only one set of results is presented (Fig. 5), where the improvement of the prediction from the proposed model seems to be quite limited. The main feature (a drop in the measured data) is not captured. In addition, the explanation of the result should be enriched. 

Response:

We added a set of experiments to validate the proposed model, processing the original data smoothly and then training and predicting. The experimental results show that the data has better performance after processing, indicating that the model has good robustness.

Comment/Suggestion 2: The prediction performance is not superior compared to other’s work, i.e., https://doi.org/10.1016/j.energy.2023.127585, which may let the reader doubts the novelty of this work.

Response:

The prediction object of this article is lead-acid batteries. Due to the more irregular attenuation curve of lead-acid batteries compared to lithium batteries, their prediction performance will be slightly inferior to that of lithium batteries. However, the existing literature on the SOH research of lead acid battery is relatively limited, so the research in this paper is still meaningful for the SOH prediction of lead acid batteries.

And we have added https://doi.org/10.1016/j.energy.2023.127585(An improved CNN-LSTM model-based state-of-health estimation approach for lithium-ion batteries) as one of the references since it is one of the latest work about state-of-health estimation approach based on LSTM model.

Comment/Suggestion 3: The SOH estimation is usually implemented up to thousands of cycles, or below 80% of the original capacity. However, it only 50 cycles with 88.1% (26/29.5) remaining. 

Response:

The prediction object of this article is lead-acid batteries. The nominal capacity of the lead-acid battery used in this article is 32 AH, and we chose 47 cyclic data for prediction, the SOH of the first cycle in the experiment is 91.6%(29.3/32), the SOH of the 47th cycle in the experiment is 81.3% (26 / 32). The amount of experimental data in this article is relatively small, and we will increase the amount of experimental data in our future research to enhance persuasiveness.

Comment/Suggestion 4: The measured degradation curve looks strange. Usually, it is a nonlinear one with an initial slowing decreasing range followed by a rapid fading one after the knee point.  It is more convincing to use publicly available experiment data, i.e., the NASA dataset (Saha B, Goebel K. Battery data set. NASA AMES prognostics data repository 2007.), Oxford dataset (Birkl C. Diagnosis and prognosis of degradation in lithium-ion batteries. University of Oxford; 2017), etc. 

Response:

The attenuation change of lead-acid battery is more complex than that of lithium battery, so the attenuation curve is more irregular compared to that of lithium battery. There are few public data sets on lead-acid batteries, and we prefer to study the health status of lead-acid batteries, more researchers have already tried the public dataset on lithium batteries.

Comment/Suggestion 5: There is not any work cited for the bat algorithm. 

Response:

We have add the related statement about bat algorithm in the section 1, and references [12-14] are cited for the bat algorithm.

Author Response File: Author Response.pdf

Reviewer 3 Report

The authors proposed a state-of-health estimation method for lead-acid batteries based on the Bat algorithm for parameter estimation. While the topic is interesting, the paper needs improvement:

1) The English requires revision.

2) Although state-of-health estimation methods for lead-acid batteries lag behind those for lithium-ion batteries, I believe it is possible to increase references and highlight the novelty of the proposed work.

3) The statement from line 107 to 110 is unclear; please revise it.

4) In line 168, one statement is repeated.

5) In section 4.1, what do you mean by 'The battery used in the study is 5 series lead-acid batteries, battery model'? And what is 'single nuclear capacity process'? Also, what is meant by 'individual battery voltage'?

6) All acronyms should be defined at their first use.

7) Unit measurements should be spaced from their numbers.

8) There are many missing spaces or extra spaces before dots.

9) In section 4.5, for instance, you stated that RMSE increased by 5.7% compared to PSO-LSTM. It is likely the contrary.

I think that moderate editing of english language is required

Author Response

Dear reviewer:

Thank you for your patient review and valuable suggestions. Here is our response to your Comments and Suggestions:

Comment/Suggestion 1: The English requires revision.

Response:

We have revised the English writing.

Comment/Suggestion 2: Although state-of-health estimation methods for lead-acid batteries lag behind those for lithium-ion batteries, I believe it is possible to increase references and highlight the novelty of the proposed work.

Response:

Some references for lead-acid batteries have been added, references [3-6,10] are related to lead-acid batteries.

Comment/Suggestion 3: The statement from line 107 to 110 is unclear; please revise it.

Response:

Thank you for your patient review and valuable suggestion, we have revised the statement.

Comment/Suggestion 4: In line 168, one statement is repeated. 

Response:

Thank you for your patient review and valuable suggestion, we have revised the statement.

Comment/Suggestion 5: In section 4.1, what do you mean by 'The battery used in the study is 5 series lead-acid batteries, battery model'? And what is 'single nuclear capacity process'? Also, what is meant by 'individual battery voltage'?

Response:

Thank you for your patient review and valuable suggestion.

The statement is wrong before，'The battery used in the study is 5 series lead-acid batteries, battery model' means ‘5 single cells for a group, and the battery type is 12 V 32 Ah’. 'single nuclear capacity process' means 'Charging and discharging process setting of a single battery 'individual battery voltage' represents the voltage of one cell for a group. We have revised the related statements.

Comment/Suggestion 6: All acronyms should be defined at their first use.

Response:

We have revised the related statements.

Comment/Suggestion 7: Unit measurements should be spaced from their numbers.

Response:

We have revised the related statements.

Comment/Suggestion 8: There are many missing spaces or extra spaces before dots.

Response:

We have revised the related statements.

Comment/Suggestion 9: In section 4.5, for instance, you stated that RMSE increased by 5.7% compared to PSO-LSTM. It is likely the contrary.

Response:

Thank you for your patient review and valuable suggestion, we have revised the related statements.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

The scientific quality of the manuscript is improved. All of my comments were addressed. 

Minor editing of English language required.

Author Response

Thank you for your patient review and valuable suggestions. Here are our responses to the Comments and Suggestions:

All the revisions are highlighted with green color.

1. We have revised the missing spaces or extra spaces before dots or missing dots, and the problem of unit measurementsalso be revised.
2. We have corrected the acronyms, such as LSTM and CCCV.
3. We have revised the description of experimental results listed in table 2 and 3 in section 4.5. In the previous version, we were not accurately expressing the information we wanted to convey, We should use “percentage improved”instead of “percentage increased”.
4. We have updated three references:[1][8][17]
5. Some revisions have been made in the introduction section.
6. We have corrected the writing mistakes.

Author Response File: Author Response.pdf

Reviewer 2 Report

Very good! All my concerns are addressed.

Author Response

Thank you for your patient review and valuable suggestions. Here are our responses to the Comments and Suggestions:

All the revisions are highlighted with green color.

1. We have revised the missing spaces or extra spaces before dots or missing dots, and the problem of unit measurementsalso be revised.
2. We have corrected the acronyms, such as LSTM and CCCV.
3. We have revised the description of experimental results listed in table 2 and 3 in section 4.5. In the previous version, we were not accurately expressing the information we wanted to convey, We should use “percentage improved”instead of “percentage increased”.
4. We have updated three references:[1][8][17]
5. Some revisions have been made in the introduction section.
6. We have corrected the writing mistakes.

Author Response File: Author Response.pdf

Reviewer 3 Report

Even though the authors improved their paper according to my previous comments, there are still some minor adjustments to do before accepting the paper.

There are still some unit measurements that should be spaced from their numbers.

Additionally, there are missing spaces or extra spaces before dots or missing dots. For instance, in line 40, the reference [1] is not spaced from the previous word. In line 130, after the word "Figure 1," a dot is missing.Some acronyms, such as LSTM, are not defined at their first use before the abstract, but they are defined later. Conversely, the acronym CCCV is defined twice. Please check.

In section 4.5, I think that the percentage increase and decrease in the comparison between the proposed model and the LSTM and PSO-LSTM are wrong. For instance, you stated that RMSE increased by 5.7% compared to PSO-LSTM. It is likely the contrary.

Minor editing of english can be required.

Author Response

Thank you for your patient review and valuable suggestions. Here are our responses to the Comments and Suggestions:

All the revisions are highlighted with green color.

1. We have revised the missing spaces or extra spaces before dots or missing dots, and the problem of unit measurementsalso be revised.
2. We have corrected the acronyms, such as LSTM and CCCV.
3. We have revised the description of experimental results listed in table 2 and 3 in section 4.5. In the previous version, we were not accurately expressing the information we wanted to convey, We should use “percentage improved”instead of “percentage increased”.
4. We have updated three references:[1][8][17]
5. Some revisions have been made in the introduction section.
6. We have corrected the writing mistakes.

Author Response File: Author Response.pdf