Review Reports

Round 1

Reviewer 1 Report

Comments for author File: Comments.pdf

Author Response

Thank you for taking the time to evaluate my work. All answers to your questions and suggestions can be found in the Word file. 

Author Response File: Author Response.docx

Reviewer 2 Report

In general, it's an interesting research work. I am just wondering the possibility for commercial application.

Author Response

Thank you for your time to review my paper and your appreciation. The developed strategy, in the end, uses neural networks, which do not require a large memory, which enables real-time applications. Moreover, the training of neural networks does not require lots of realistic collected driving data due to the utilization of data expansion technology. Finally, energy management using the LSTM network involves inputs about the average power, SoC, load power demand, and the specific consumption curves of the fuel cell system, and this information is available. 

Reviewer 3 Report

the article called "Machine Learning-based Control for Fuel Cell Hybrid Buses: from Average Load Power Prediction to Energy Management" is worthy of publication. I think it is mature and tackles all the issues fully (sometimes too much).

The introduction is rich, very complete. The model can then be followed in the various steps.

I would have liked to see in figure 1 a definition of the powers of the various parts. If the system has a power section to be used for air conditioning. I think that what has been addressed (it is excellent) does not also take into account seasonal variability, considering the heating/cooling needs of the buses. I do not pretend that these changes are done, the article is already good, I would like to suggest for further development.

I liked the aging analysis.

For me the article can find publication, congratulations.

Author Response

Thank you for your time to review my paper and your appreciation.  I change the figure 1 to add the definitions of various power as follows. Regarding the auxiliary power consumption, the heating/cooling power strongly depends on the seasons. In this work, it is assumed constant. Therefore, the forward neural network of estimating the average fuel cell power can be improved by adding another season time input in the future.

Author Response File: Author Response.docx