Next Article in Journal
CFD Simulations of Snowdrifts on a Gable Roof: Impacts of Wind Velocity and Snowfall Intensity
Next Article in Special Issue
Energy Cost Driven Heating Control with Reinforcement Learning
Previous Article in Journal
Numerical Simulation of Bearing Characteristics of Bored Piles in Mudstone Based on Zoning Assignment of Soil around Piles
Previous Article in Special Issue
Hybrid Model for Forecasting Indoor CO2 Concentration
 
 
Article
Peer-Review Record

A Model Predictive Control for Heat Supply at Building Thermal Inlet Based on Data-Driven Model

Buildings 2022, 12(11), 1879; https://doi.org/10.3390/buildings12111879
by Liangdong Ma, Yangyang Huang, Jiyi Zhang and Tianyi Zhao *
Reviewer 1: Anonymous
Reviewer 2: Anonymous
Reviewer 3:
Reviewer 4:
Reviewer 5:
Buildings 2022, 12(11), 1879; https://doi.org/10.3390/buildings12111879
Submission received: 6 September 2022 / Revised: 17 October 2022 / Accepted: 1 November 2022 / Published: 4 November 2022

Round 1

Reviewer 1 Report

Authors have present an interesting research paper on the topic of MPC for building energy systems. Some major issues are as follows:

1. The data description is not very clear. THe sensors localtions and their uncertainty are not clearly present. Especially Table 2, it would be very difficult to understand the meanings of sensor 'building B' and 'building C'.

2. The research gap between this MPC study and the previous MPC studies is not clear.

3. Some practical application issues. It is suggested that authors should add some practical application issues about the applications of MPC for buidling energy saving goals.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The paper is interesting as it addresses some of the contemporary topics in advanced HVAC controlMPC is getting increasingly popular in the research community, although its commercial applications are still relatively rare. Also the use of data-driven models is of high interest for today's practitionersThis research paper does not address a big challenge (since optimal control of one or two parameters in a district heating system is not a complex problem) but it provides a useful case study.

 

In the introduction, line 64, I agree with forward modeling and data-driven modeling, but I have two comments:

1) Should not be the word "forward" replaced by "first principle" modeling? I think the word "forward" is confusing

2) Also it would be good to mention standard state-space models, which are determined by identification of the model from dynamic data, typically based on step-testing. I believe this type of model is far most typical for MPC applications.

Probably the major gap in the paper I see in section 4, which captures the control performance analysis. The impact of LSTM and PSO is analyzed using "simulation" (e.g. section 4.2) but it is not explained how was the simulation model created. Were three different simulation models created for all three buildings? And what type of simulation models were applied? I originally thought the proposed methods were applied directly to control the physical district system but if there was a simulation used instead, it should be explained the nature of the simulation model.

Some sentences are difficult to read and the whole text should be reviewed and corrected. From some reason, the words "data-driven" is used but the noun is missing - for instance, the title in the current form makes little sense I guess in many cases it should be "data-driven strategy" or "data-driven model".

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

The article should be rejected for the following reasons:

1.      The methods are not explained clearly. For instance, there is no explanation on training and testing data. What you call "subspace model identification" would require a chapter alone to describe the underlying physical model, the disturbance signals used etc. There is no information about this, which makes statements such as those on the comparison between SMI and LSTM not scientifically sound.

2.      There is no discussion on the main limitation of the proposed approach:1) PSO does not guarantee optimal solutions and is rather slow, 2) LSTM do not guarantee physical consistency especially when boundary conditions change.

3.      By reading the title and the abstract, it looks like experimental tests were performed. In fact, this study is based on simulations.

4.      "MPC based on data-driven" does not mean anything, because data-driven is like an adjective.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 4 Report

You have presented your work in the form of good practice.

I have no comments. I wish you success in continuing the work

Author Response

Thanks for your comments and suggestions. Hope you have a good day!

Reviewer 5 Report

In this manuscript, the authors proposed a method for predicting indoor thermal environment and conducted a study to apply it to control. The method proposed by the authors accurately predicted the indoor thermal environment, and when applied to control, a good effect could be expected. It contains good content, but the reviewer's comments are as follows.

 

1. The prediction accuracy of the indoor thermal environment was quantitatively evaluated using the coefficient of determination (R2), but the quantitative evaluation of the control performance was insufficient. Evaluate control performance by introducing quantitative indicators such as time constant or overshoot of control.

 

2. Indoor thermal environment or thermal comfort is very sensitive to temperature. From that point of view, measurement using a temperature sensor with a precision of 0.5°C seems controversial. It is thought that a more precise temperature measurement was necessary through a separate calibration.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Back to TopTop