A Deep-Learning-Based Meta-Modeling Workflow for Thermal Load Forecasting in Buildings: Method and a Case Study
Round 1
Reviewer 1 Report
The assessed work addresses issues related to forecasting cooling and heating loads in buildings at an early design stage. The authors used meta-models to predict the heating and cooling demand. Load profiles for an office building, a retail building and a hotel obtained from EnergyPlus were used to build the model. The model was verified on buildings located in the western district of Hongqiao in Shanghai.
The work is very interesting and has great practical significance. It is prepared very well. I have included minor comments in the paper.
Before its publication, I suggest:
1. Expand the price / criticism of known methods. I find the statement that methods based on building physics are complicated and time consuming insufficient. Maybe refer to works such as e.g. https://doi.org/10.1016/j.enbuild.2021.111160.
2. What has guided the selection of variables for the model? The paper [doi:10.26868/25222708.2019.210436.] only deals with office buildings. It contains sets of 7 to 13 variables. How might the choice of variables affect the outcome of the modelling?
3. In Figure 10 - in what units is frequency expressed?
4. Modify the notation of equation 2 so that the result is in %.
5. Table 3 - what is the reason for the increase in MAPE error for cooling? Wrong set of input variables?
6. Figure 15 - how was the Relative Error determined? Missing from the methodology.
7. No reference of own research results to literature.
8. Lack of limitations and follow-up plans by the authors.
Overall, I rate the paper well and look forward to the final version.
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 2 Report
Please check attached file
Comments for author File: Comments.pdf
Author Response
Please see the attachment.
Author Response File: Author Response.pdf