Which price to pay? Auto-tuning building MPC controller for optimal economic cost
Jiarui Yu, Jicheng Shi, Wenjie Xu, Colin N. Jones · 2025
Read Which price to pay? Auto-tuning building MPC controller for optimal economic cost by Jiarui Yu, Jicheng Shi, Wenjie Xu, Colin N. Jones (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2501-10859.txt.
Introduction
Jiarui Yu, Jicheng Shi, Wenjie Xu, and Colin Jones wrote this 2025 paper on tuning a building's smart heating and cooling controller to cut power bills. Pricing plans that reward flexible use are complex, and forecast-based controllers can exploit them, but only when their settings are tuned well. The team automates that tuning with a guarded form of Bayesian search.
In a case study, the tuned controller cut power costs by about twenty-seven percent against a rule-based one. It also beat a hand-tuned version of the same controller by about seventeen percent. A scan of real price contracts showed that picking the best plan alone could trim monthly bills by up to about twenty percent. The authors present the pairing of smart control and smart plan choice as a data-driven route to savings.
Catalog record
| CREATOR | Jiarui Yu, Jicheng Shi, Wenjie Xu, Colin N. Jones |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 59 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2501-10859
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Which price to pay? Auto-tuning building MPC controller for optimal economic cost Jiarui Yu, Jicheng Shi, Wenjie Xu, Colin N. Jones Abstract Increasing types of price-based demand-side management (DSM) programs offered in the residential household market presents a challenge for consumers to select the most cost-effective option. In this paper, we aim to offer guidance to the consumers based on simulation statistics which considers the optimal costs of possible programs in the market and requirements from the consumers. Model predictive control (MPC) controller was considered for temperature management in buildings but its performance heavily depends on hyperparameters. Consequently, MPC necessitates meticulous hyperparameter tuning to attain optimal performance under diverse contracts. However, conventional building controller design is an open-loop process without critical hyperparameter optimization, often leading to suboptimal performance due to unexpected environmental disturbances and modeling errors. Furthermore, these hyperparameters are not adapted to different pricing schemes and may lead to non-economic operations. To address these issues, we propose an efficient performance-oriented building MPC controller tuning method based on a cutting-edge efficient constrained Bayesian optimization algorithm, CONFIG, with global optimality guarantees. We demonstrate that this technique can be applied to efficiently deal with real-world DSM program selection problems under customized black-box constraints and objectives. To substantiate the efficacy of the proposed approach, a case study was conducted on the building simulation platform, BOPTEST. In this study, a simple MPC controller, which offers the advantages of reduced commissioning costs, enhanced computational efficiency, was optimized to perform on a comparable level to a delicately designed and computationally expensive MPC controller. The results also indicate that with an optimized simple MPC, the monthly electricity cost of a household can be reduced by up to 26.90% compared with the cost when controlled by a basic rule-based controller under the same constraints. Then we compared 12 real electricity contracts in Belgium for a household family with customized black-box occupant comfort constraints. The results indicate a monthly electricity bill saving up to 20.18% when the most economic contract is compared with the worst one, which again illustrates the significance of choosing a proper electricity contract. † † graphicalabstract: † † highlights: Constrained Bayesian Optimization of MPC controller in building control. An efficient quadratic programming MPC for a case study of one month. Optimized quadratic programming MPC achieves better performance than a complex dedicately designed MPC with much lower commissioning costs. Offering guidance for customers with 12 different real electricity contracts under customized black-box constraints Up to 20.18% monthly saving with the most economic contract compared with the worst contract. 1 Introduction Nomenclature Abbreviations ANN Artificial Neural Network ARX AutoRegressive model with eXogenous inputs BO Bayesian Optimization CDF Cumulative Distribution Function CONFIG Constrained Efficient Global Optimization DSM Demand-side Management GA Genetic Algorithm HVAC Heating, Ventilation, and Air Conditioning i.i.d Independent and identically distributed MIQP Mix Integer Quadratic Programming ML Machine Learning MPC Model Predictive Control NN Neural Network PMV Predicted Mean Vote RL Reinforcement Learning Variables \beta^{1/2} Weighting parameter in CONFIG g Closed-loop performance constraint \epsilon Slack variable for temperature constraints \lambda Gaussian process noise variance \mathcal{B} Total number of DSM programs to choose from \mathcal{U} Constraint set for control input \mathcal{Y} Constraint set for indoor temperature J Closed-loop performance objective \Theta Sample sequence \theta Tuning parameters in CONFIG algorithm A(q) Coefficient for previous output K Total number of iterations in CONFIG N Total time steps R Linear penalty coefficient for heating power S Penalty coefficient for slack variable t_{d} Time delay u Input heat pump power y Indoor Temperature y_{ref} Temperature setpoint Subscripts 0 Measurement of objective in CONFIG 1 Measurement of constraint in CONFIG config For CONFIG problem k Iteration steps for CONFIG mpc For MPC problem t Index of time steps for building model Rapid urbanization and population growth have significantly increased global energy consumption and carbon emissions, placing immense pressure on the environment and natural resources. Buildings alone account for approximately 34% of global energy use 20.500.11822_41133 , with nearly half of this consumption attributed to HVAC systems Benchmark . Optimizing the control of these systems offers a substantial opportunity to enhance energy efficiency, with research indicating potential reductions in energy usage of up to 30% LIU2023113436 . Price-based demand-side management (DSM) programs, such as time-of-use (TOU) pricing, real-time pricing (RTP), and critical peak pricing (CPP), aim to alleviate this burden by incentivizing consumers to adjust their energy usage according to fluctuating electricity prices. However, the DSM programs present additional challenges for consumers and building operators as they face a wide range of electricity billing contracts with varying rates, peak pricing structures, and time-based tariffs EID201615 . This complexity makes it difficult to select the most cost-effective contract, which is further complicated by the necessity of optimizing building control strategies to minimize electricity bills while maintaining comfort. Indeed, traditional rule-based control methods frequently prove inadequate for the effective management of these dynamic and diverse pricing schemes YANG2022120054 , YANG2020115147 . Model Predictive Control (MPC) has emerged as a promising solution for addressing these challenges due to its predictive and flexible nature CIGLER201239 , TABARESVELASCO20191346 . MPC predicts future electricity prices and building dynamics, allowing for proactive optimization of control actions in a receding-horizon manner. This capability leads to superior performance and improved robustness for managing buildings under various billing contracts 5717458 , MBUNGU20173419 , YANG2022120054 . For instance, Yang et al. YANG2022120054 developed an MPC controller considering a highly dynamic electricity prices for demand-responsive building control electricity prices for demand-responsive building control. When compared to two other conventional controllers, the MPC controller demonstrated a reduction in heating energy consumption of up to 13.8% and a reduction in electricity bill of up to 70%, while maintaining an acceptable level of comfort. However, the effectiveness of MPC depends heavily on the choice of cost and constraint functions and other parameters such as model structure and coefficients, setup of constraints and objectives, to achieve desired performance. The traditional approach to MPC design often consists of two separate stages PAN2023100135 : building model construction and calibration, and controller design. Regarding the modeling step, significant methods have been explored for building modeling with accurate prediction, including Resistor-Capacitor models WANG201951 , Behavior methods 10089206 , Gaussian Processes MASSAGRAY2016119 , and Neural Networks FAN2019700 , HU2021117429 , YANG2023100144 . In addition, Bayesian calibration CALAMAGONZALEZ2021116118 , MARTINEZVIOL2022109693 , ZHAN2022112278 , CHAKRABARTY2021111460 can be used for further improvement. Based on the developed models, MPC controllers have been developed for various electricity billing contracts YANG2022120054 , stochastic environments Oldewurtelinproceedings , dynamic estimates and predictions JOE201965 . It should be noted that the two-stage focus on the open-loop optimality separately, that is the modeling prediction accuracy and open-loop objective and constraint satisfaction. However, the resulting MPC may fail to achieve overall optimal performance in the closed-loop control, which also suffers from various disturbances and modeling errors due to complex components in buildings. Given the above suboptimal performance of MPC controller in closed-loop operation, an inclusive hyperparameter optimization is necessary to guide consumers in selecting the most cost-effective billing policy. For existing companies offering contract selection guidance ENGIEhelp , selectandswitch , historical data is collected to predict electricity usage patterns su15118750 , and the most suitable billing contract is chosen based on this simulated usage. However,