Time-Series Prediction of Electricity Load for Charging Piles in a Region of China Based on Broad Learning System

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper focuses on predicting the electricity load in China using key factors that affect it. The literature review for time series forecasting is a bit lacking. The efficacy of the work is demonstrated through data examples and compared against other methods but the experiment setting is a bit unclear.

 

Below are my comments:

 

·       There is a typo in the abstract “Finallyparticle”. Similarly in line 139, “factorswhich”. This is a common occurrence across the paper. Please address this.

·       What is f(t) in equation 2?

·       Is there any particular reason PSO was used instead of Genetic Algorithm or any other method?

·       Can the authors please expand on the dataset information? What does 460 sets mean? It is 460 days and how many timepoints are there in each day?

·       I might have missed something, but please clarify this for the authors. In the current data, was the historical value of the electricity load the only feature used for the data analysis? If not, can you share information about the other features? Also, in Fig 1, it talks about Climate factors and Random factors, how was that incorporated into the experiment?

·       Also, can the author share the experimental setup for LSTM? What exactly was the input and output?

·       Also, it would be useful to share the model information to understand how deep the models were, etc.

·       How is overfitting handed for the methods?

·       Can the authors add simple Arima as a base comparison?

·       It would be great if the authors added a few lines in the conclusion for future work.

·       For Fig3-7, the residual plot can be removed as the real vs predicted plot gives that information along with Table 1. This will make the paper more concise.

·       Please proofread the paper and remove the typos.

·       Literature for time series prediction can be enriched by adding this paper. https://link.springer.com/article/10.1007/s11222-023-10299-z

Comments on the Quality of English Language

Please check above

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

1.       Add a brief quantitative analysis of the advantages of the proposed method over previous methods at the end of the conclusion in the abstract section of the paper for parameter values ​​i.e. R², P_RMSE, P_MAE, and P_MAPE.

2.       Figure 1 is not readable, please revise and colour the image block so that it is brighter and more striking so that it can attract readers' interest in understanding the meaning of this figure.

3.       There are many abbreviations in this paper, please create a new table containing the abbreviations and their meanings from the beginning and end of the paper.

4.       This paper has many equation symbols, please create a new table containing the equation symbols and explain their meaning from the beginning and end of the paper.

5.       The author mentions the Broad Learning System (BLS) as the method proposed in the title and discussion of the paper to determine Time-Series Prediction of Electricity Load (see manuscript title). However, in the comparative analysis of results, the BLS method has been used by another researcher (See Table 1 and Reference 22). The question is what is the existing name of the method proposed by the author? If the term proposed for the author method is different (not BLS), it is best to create a new method name to differentiate it from the BLS method by other researchers in Reference 22.

 

6.       In the conclusion section, explain the weaknesses of your proposed method and explain future work that can be proposed to resolve these weaknesses. 

Comments for author File: Comments.pdf

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

This is a good piece of work and I appreciate the authors for their work and achived results. However I some comments on Table 1.

1. Do you think that P_RMSE and P_MAE are accurate judge for your proposal, since these two criteria cannot be used to compare the prediction between different time-series.

2. The table showed diffrent criteria for the sake of the roposed model with other different models, e.g.. LMSTM, NARX, etc. The question is; why you used models that implemented in different applications, rather than forcasting charging load. The Refrences [21-23], their applications are different. You may compare your proposed model with LSTM model that used for charging load forcasting [10.25236/AJETS.2022.051301] and the result of MAPE% was good (4.17).. Please explain.

Comments on the Quality of English Language

The are some typo and grammar issue. for example:

In Abstract;  Finallyparticle swarm optimization ..... should be corrected to; Finally, particle swarm optimization ....

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for addressing the comments and for the explanantions.

 

I still have two comments which I would appreciate clarification on:

 

1. Can the authors please expand on the dataset information? What does 460 sets mean? It is 460 days and how many time points are there in each day? The author's answer was not clear.

 

2. How is overfitting handled for the methods? Can the authors add simple Arima as a base comparison? The author's answer was not clear. Adding Arima should be a quick task to get a baseline comparison.

Comments on the Quality of English Language

Please proofread the paper

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The revision from the authors is enough.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf