A sensor-driven physics-informed neural network for real-time occupant-centric space heating control
Organizations: School of Engineering, University of Warwick, Coventry CV4 7AL, United Kingdom · McMaster Automotive Resource Centre, McMaster University, Hamilton, ON L8P 0A6, Canada
Abstract
Conventional space heating systems typically rely on single or sparse fixed-point temperature measurements, overlooking non-uniform indoor temperature distributions, transient thermal dynamics, and occupant-specific comfort requirements. This study explores an occupant-centric control strategy that aims to maintain the desired temperature around the occupant location. Spatially and temporally varying temperature fields, together with occupant mobility, make real-time local temperature estimation challenging. To address this challenge, we propose a control-oriented Physics-Informed Neural Network (PINN) model for indoor temperature-field prediction. Conventional PINNs rely on explicit time inputs and assume consistent temporal patterns across samples, which limits adaptability under variable initial conditions and heating sequences. The proposed model infers temporal dynamics implicitly from spatial coordinates and limited sensor measurements, improving robustness to changes in initial conditions and external disturbances. The method was validated across nine horizontal sensing planes in three controlled room-scale configurations. Against a validated COMSOL Multiphysics simulation, the PINN predicted the evaluated temperature fields with a maximum relative error below 1.85% across validation cases. These predictions supported an occupant-centric space heating control scheme. Compared with COMSOL-based offline control, the PINN-based system achieved an average prediction error of around 0.67%, and its rapid inference supports real-time occupant-centric control implementation. This work represents a move away from reliance on fixed-point temperature measurements in space heating control to an occupant-centric control strategy informed by real-time environmental feedback.
Figures & tables
| No. | Hyperparameters | Values |
|---|---|---|
| 1 | Number of neurons in input layer of Branch 1 | 2 |
| 2 | Number of neurons in input layer of Branch 2 | 9 |
| 3 | Number of neurons in the output layer | 1 |
| 4 | Number of neurons in fully connected layers of Branch 1 | 512 |
| 5 | Number of neurons in fully connected layers of Branch 2 | 512 |
| 6 | Size of the hidden layer after fusion | 4 |
| Cases Error | 1kW | 2kW | 3kW | 3kW 2kW | 3kW 1kW | 2kW 1kW |
|---|---|---|---|---|---|---|
| Maximum relative error (%) | 2.69 | 4.60 | 4.55 | 4.24 | 2.87 | 2.53 |
| Average relative error (%) | 0.42 | 0.80 | 0.94 | 0.80 | 0.50 | 0.51 |
| Average absolute error ( ) | 0.090 | 0.176 | 0.194 | 0.164 | 0.105 | 0.106 |
| Standard deviation ( ) | 0.097 | 0.134 | 0.144 | 0.131 | 0.091 | 0.083 |
| Cases Error | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 |
|---|---|---|---|---|---|---|---|---|---|---|
| Maximum relative error (%) | 1.60 | 1.69 | 1.09 | 1.33 | 1.83 | 1.68 | 1.71 | 1.50 | 1.84 | 1.75 |
| Average relative error (%) | 1.13 | 0.78 | 0.63 | 0.61 | 0.81 | 0.78 | 1.14 | 0.61 | 0.77 | 0.77 |
| Average absolute error ( ) | 0.243 | 0.166 | 0.135 | 0.133 | 0.177 | 0.170 | 0.243 | 0.131 | 0.165 | 0.162 |
| Standard deviation ( ) | 0.043 | 0.053 | 0.037 | 0.042 | 0.051 | 0.057 | 0.044 | 0.046 | 0.041 | 0.050 |
Heating, Ventilation, and Air Conditioning Control
Heating, ventilation, and air conditioning control optimizes building energy use while maintaining thermal comfort, indoor air quality, and equipment safety.
See all 27 papers