• No results found

In accordance with the optimization framework and parameterization methods pre-sented, future work can readily build upon the design to incorporate optimization strategies focused on energy saving and output controls for the heating system along with the windows. When updating constraints, future work may additionally factor in the time it takes the windows to open and close, about 1 second per 1% increase or decrease in opening extent.

Directly from this design contribution, a foreseeable next step is to build the software system interface for real-time control implementation and model develop-ment. With this set-up, the model may be built using optimized parameterization algorithms that better handle data gaps in the past several hours of server data, handling noise as necessary. The MPC algorithm may use that model to predict and implement controls within the next hour, even as the next model is trained incorporating room responses to those controls. This will transform the project from predictive to adaptive control, as the controller adapts its predictions based on up-to-date models [29]. Software integration will also be necessary to accommodate larger and more complex spaces within the whole building of HouseZero, and other building applications.

Real-time implementation will not only bring validation data closer in time to the training data, as compared to using a November 2018 model on March 2019 data.

It will also increase the frequency of updates in disturbance variables. Limited to 5 minute intervals during the control trials by manual implementation, updates to disturbance inputs may happen on the scale of seconds to a minute in the future software interface. These factors will likely contribute to a higher prediction accu-racy for the MPC algorithms, driving achievement of ambitious energy and thermal performance goals.

Beyond predictive and adaptive MPC, it will be fascinating to see future work with other system identification methods and control design approaches, branching into artificial neural networks and machine learning algorithms with the wealth of data available from building and city sensor networks.

Bibliography

[1] ASHRAE, “Ventilation and Infiltration,” in Handbook of Fundamentals, Inch-Pound, Ed. Atlanta, GA: ASHRAE, 2017, pp. 16.1-16.32.

[2] ASHRAE, “Energy Estimating and Modeling Methods,” in Handbook of Funda-mentals, Inch-Pound Ed. Atlanta, GA: ASHRAE, 2017, pp. 19.1-19.42.

[3] ˚Astrom, K.J., & Murray, R.M. “PID Control” in Feedback Systems: An Introduc-tion for Scientists and Engineers, Second Ed., Princeton University Press, 2019, pp. 2-27.

[4] Atkinson, J., Chartier, Y., Pessoa-Silva, C.L., et al., editors. “Understanding natural ventilation,” Natural Ventilation for Infection Control in Health-Care Settings. Geneva: World Health Organization, 2009. Available from: https:

//www.ncbi.nlm.nih.gov/books/NBK143284/. [Accessed Mar. 31, 2019].

[5] Beck, A. & Guttam-Beck, N. “FOM – A MATLAB Toolbox of First Order Methods for Solving Convex Optimization Problems,” in Optimization Methods and Software, 2018.

[6] Bosschaerts, W. Renterghem, T.V., Hasan, A., O., & Limam, K. “Development of a model based predictive control system for heating buildings,” Energy Procedia 112, 519-528, 2017.

[7] Boyd, S. & Vandenberghe, L. Convex Optimization. Cambridge UK: Cambridge University Press, 2004.

[8] Buonomano, A., Montanaro, U., Palombo, A., & Santini, S. “Dynamic build-ing energy performance analysis: A new adaptive control strategy for strbuild-ingent thermohygrometric indoor air requirements,” Applied Energy, 163, pp. 361-386, 2016.

[9] da Costa Sousa, J.M. & Kaymak, U. “Model predictive control using fuzzy de-cision functions,” IEEE Transactions on Systems, Man, and Cybernetics–Part B:

Cybernetics, 31, pp. 54-65, 2001.

[10] Dorf, A., Fellman, C., Dokka, T.H., Lassen, N., Myrup, M, Fjellheim, H., Malkawi, A., et al. House Zero Report. Cambridge, MA: Harvard Center for Green Buildings and Cities, 2018.

[11] Flint, T.W., & Vaccaro, R.J. “Performance Analysis of N4SID State-Space Sys-tem Identification,” in Proceedings of the American Control Conference, Philadel-phia, PA, 1988.

48

[12] Friedland, B. Control System Design – An Introduction to State-Space Methods.

Dover, Ed. New York: McGraw-Hill, 1986.

[13] Gunantara, N. “A review of multi-objective optimization: Methods and its ap-plications,” Cogent Engineering, 5, 1502252, 2018.

[14] Harvard Center for Green Buildings and Cities. “CGBC Headquarters: House-Zero.” [Online]. Avaliable: http://harvardcgbc.org/research/housezero/.

[Accessed Apr. 2, 2019].

[15] Hachicha, S., Kharrat, M., & Chaari, A. “N4SID and MOESP Algorithms to Highlight the Ill-conditioning into Subspace Identification,” International Journal of Automation and Computing, 11(1), 30-38, 2014.

[16] Hu, J., & Karava, P. “A state-space modeling approach and multi-level opti-mization algorithm for predictive control of multi-zone buildings with mixed-mode cooling,” Building and Environment, 80, 259-273, 2014.

[17] Katbab, A. “Fuzzy logic and controller design: a review,” in Proceedings IEEE Southeastcon ’95. Visualize the Future, Raleigh, NC, USA, Mar. 1995.

[18] Li, S., Lim, K., & Fisher, D. “A state space formulation for model predictive control,” AIChE Journal, 35(2), 241-249, Feb. 1989.

[19] Li, S., Joe, J., Hu, J., & Karava, P. “System identification and model-predictive control of office buildings with integrated photovoltaic-thermal collectors, radiant floor heating and active thermal storage,” Solar Energy, 113, 139-157, 2015.

[20] Ljung, L. “Introduction,” System Identification: Theory for the User, Second Ed., Upper Saddle River, NJ: Prentice-Hall, 1999.

[21] Lundh, M. & Molander, M. “State-Space Models in Model Predic-tive Control,” ABB Automation Products AB, n.d. [Online]. Available:

https://library.e.abb.com/public/c8687aec93abed2e85256f9b0054e8fb/

TP_Lundh_Molander.pdf. [Accessed Mar. 31, 2019].

[22] MathWorks Documentation. “Compare and Lsim” The MathWorks, Inc., 2018.

[23] Montgomery, D.C., Peck, E.A., & Vining, G.G. Introduction to linear regression analysis. Wiley series in probability and statistics, 5th ed. Hoboken, N.J.: Wiley, 2012.

[24] Nesterov, Y. Introductory Lectures on Convex Optimization: A Basic Course, Springer Science+Business Media, LLC, Applied Optimization, 87, 2004.

[25] Nguyen, D.J. & Widrow, B. “Neural networks for self-learning control systems,”

IEEE Control Systems Magazine, pp. 18-23, 1990.

[26] Oldewurtel, F., Parision, A., Jones, C.N., Gyalistras, D., Gwerder, M., Stauch, V., Lehmann, B., & Morari, M. “Use of model predictive control and weather forecasts for energy efficient building climate control,” Energy and Buildings, 45, pp. 15-27, 2012.

Data-Driven Model Development for Model Predictive Building Control

[27] RuppAir. Actual Air Density Calculation. RuppAir, (n.d). Available from https://www.ruppair.com/documents/white-papers/Actual%20Air%

20Density%20BTU%20Calculation.pdf. [Accessed Mar. 31, 2019].

[28] Salem, F. & Mosaad, M.I. “A comparison between MPC and optimal PIC con-trollers: Case studies,” in Michael Faraday IET International Summit, Kolkata, India, Sep. 2015.

[29] Shaikh, P.H., Nor, N.B.M., Nallagownden, P., Elamvazuthi, I., & Ibrahim, T.

“A review on optimized control systems for building energy and comfort manage-ment of smart sustainable buildings,” Renewable and Sustainable Energy Reviews, 34(C), pp. 409-429, 2014.

[30] United Nations Environment and International Energy Agency. “Towards a zero-emission, efficient, and resilient buildings and construction sector: Global Status Report 2017,” United Nations Environment Programme, 188, 2017.

[31] Yan, B. “Control Design for Natural Ventilation and Thermally Activated Building System in a Zero Energy Building,” Forthcoming, 2019.

[32] Yang, S., Wan, M., Ng, B., Zhang, T., Babu, S., Zhang, Z., Chen, W., and Dubey, S. “A State-space Thermal Model Incorporating Humidity and Ther-mal Comfort for Model Predictive Control in Buildings,” Energy and Buildings, 170(25), 2018.

[33] Yao, Y., Yang, K., Huang, M., & Wang, L. “A state-space model for dynamic response of indoor air temperature and humidity,” Building And Environment, 64, 26-37, 2013.

[34] Zhang, R., Wu, S., & Gao, F. “State Space Model Predictive Control for Ad-vanced Process Operation: A Review of Recent Development, New Results, and Insight,” Industrial & Engineering Chemistry Research, 56(18), 5360-5394, 2017.

50 Chapter Ald´ıs Elfarsd´ottir

Appendix

A.1 System matrices, third iteration models

Fullrow-2nd Order Atrain-2nd Order Dimensions A coefficients state 1, state 2 state 1, state 2 2x2 state 1 0.9984, -0.0378 0.9985, -0.0562

state 2 1.5630e-4, 0.1769 0.0115, 0.3484

C coefficients state 1, state 2 state 1, state 2 1x2 room temp -136.1071, 1.1693 41.7302, -0.9131

K coefficients room temp room temp 2x1

state 1 -0.0089 0.0349

state 2 0.0337 -0.0730

Table A.1: A,C,K coefficients for states, response, and prediction error

Fullrow-2nd Order Atrain-2nd Order Dimensions B coefficients state 1, state 2 state 1, state 2 2x1 positR (%) -3e-6, -1e-4 2e-7, 2e-6

F coefficients state 1, state 2 state 1, state 2 2x13 positS (%) -4e-7, 8e-6 -1e-7, -4e-7

heatvalve (%) 4e-7, -1e-5 -2e-7, 6e-6 slab (oC) 1e-2, 1e-1 -5e-3, -1e-1 humidity (%) 2e-6, 1e-5 -7e-8, -4e-6 co2 (ppm) 5e-7, 4e-6 -5e-7, -1e-5

occupancy 2e-4, -9e-3 -8e-6, 5e-4

wind direct (o) 1e-7, 9e-7 -2e-9, 3e-7

rain -1e-6, -2e-4 9e-7, 4e-5

outdoor (oC) -2e-5, -3e-4 -5e-7, 2e-8 temp31 (oC) -4e-3, -5e-2 -6e-4, -1e-2 temp33 (oC) 7e-4, 8e-3 -1e-4, -3e-3 temp23 (oC) -3e-5, -3e-4 -4e-5, -8e-4 temp22 (oC) -6e-4, -7e-3 -8e-5, -2e-3

Table A.2: B,F coefficients for control and disturbances

Data-Driven Model Development for Model Predictive Building Control

Related documents