IMPROVING BUILDING EQUIPMENT ENERGY EFFICIENCY BASED ON PARTIALLY OBSERVABLE MARKOV MODEL
Journal of Green Building · 2026 · 21 (3) · 83-106
DOI: 10.3992/jgb.21.3.83 ↗At a glance
Journal article in Journal of Green Building, 2026. Written by Heer Guo. 37 references.
Abstract (original language)
This study aims to enhance the energy efficiency of building equipment using a Partially Observable Markov Decision Process (POMDP) model as the analytical tool. By modeling partially observable states in HVAC operation, the approach yields adaptive maintenance strategies based on state estimates, optimizing energy use across conditions. The implementation of these adaptive strategies is expected to simultaneously optimize equipment performance and minimize energy consumption, thereby directly contributing to sustainable development objectives. Case simulations demonstrate the effectiveness of this POMDP strategy, which achieved a minimum convergence cost of 143.149 USD (convergence cost refers to the expected long-term total system cost after the optimal POMDP decision policy stabilizes) and reduced maintenance costs by over tenfold compared to fixed sensor inspection schedules. Furthermore, the solution converged rapidly, within ∼1000 seconds using the SARSOP algorithm. The results of this research provide theoretical and methodological support for enhancing the energy efficiency of building energy equipment, holding practical significance in promoting green building practices and sustainable energy.
Cite
APA
Guo, H. (2026). IMPROVING BUILDING EQUIPMENT ENERGY EFFICIENCY BASED ON PARTIALLY OBSERVABLE MARKOV MODEL. Journal of Green Building, 21(3), 83-106. https://doi.org/10.3992/jgb.21.3.83
Vancouver
Guo H. IMPROVING BUILDING EQUIPMENT ENERGY EFFICIENCY BASED ON PARTIALLY OBSERVABLE MARKOV MODEL. Journal of Green Building. 2026;21(3):83-106. doi:10.3992/jgb.21.3.83
BibTeX
@article{Guo2026,
author = {Guo, Heer},
title = {IMPROVING BUILDING EQUIPMENT ENERGY EFFICIENCY BASED ON PARTIALLY OBSERVABLE MARKOV MODEL},
journal = {Journal of Green Building},
year = {2026},
volume = {21},
number = {3},
pages = {83--106},
doi = {10.3992/jgb.21.3.83}
}
EUVIMED