What question did this study set out to answer?

To develop and evaluate a physics-informed neural network-based model predictive control (NN-MPC) framework that co-optimizes four residential energy assets—photovoltaics, battery storage, heat-pump HVAC, and electric vehicles—under time-of-use pricing and real-time carbon intensity.

August 14, 2026Journal of Building Performance Simulation

Neural network-based model predictive control for residential four-asset energy co-optimization with vehicle-to-home integration and carbon-aware dispatch

View Full Paper

Ask AI

Bookmark

Authors

SASellami Ali BDBenlahcene Djaouida

Discussion

Member takes

Overview

Simulation study demonstrates that neural network model predictive control cuts residential emissions and electricity costs, indicating scalable real-time multi-asset smart home optimization.

Key Points

To develop and evaluate a physics-informed neural network-based model predictive control (NN-MPC) framework that co-optimizes four residential energy assets—photovoltaics, battery storage, heat-pump HVAC, and electric vehicles—under time-of-use pricing and real-time carbon intensity.
Trained an offline feedforward neural network with a cost-augmented physics loss function to replace the online receding-horizon optimization solver.
Evaluated performance using full-year Monte Carlo simulations (30 runs) with Algiers climate data, benchmarked against rule-based and mixed-integer linear programming (MILP) controllers.
NN-MPC reduced thermal comfort deviation by 50.8%, electricity costs by 7.5%, grid consumption by 21.4%, and carbon emissions by 26.3% compared to standard control.
Achieved sub-millisecond inference times, recovering 87% of the MILP optimality gap at over 2,400 times lower computational cost.

Cite This Study

Ali et al. (2026) studied this question.

synapsesocial.com/papers/6a7ec707b70b84ec8b913254 https://doi.org/10.1080/19401493.2026.2712297

View Full Paper

Ask AI

Bookmark

August 14, 2026Journal of Building Performance Simulation

Neural network-based model predictive control for residential four-asset energy co-optimization with vehicle-to-home integration and carbon-aware dispatch

View Full Paper

Ask AI

Bookmark

Authors

SASellami Ali BDBenlahcene Djaouida

Discussion

Member takes

Overview

Simulation study demonstrates that neural network model predictive control cuts residential emissions and electricity costs, indicating scalable real-time multi-asset smart home optimization.

Key Points

To develop and evaluate a physics-informed neural network-based model predictive control (NN-MPC) framework that co-optimizes four residential energy assets—photovoltaics, battery storage, heat-pump HVAC, and electric vehicles—under time-of-use pricing and real-time carbon intensity.
Trained an offline feedforward neural network with a cost-augmented physics loss function to replace the online receding-horizon optimization solver.
Evaluated performance using full-year Monte Carlo simulations (30 runs) with Algiers climate data, benchmarked against rule-based and mixed-integer linear programming (MILP) controllers.
NN-MPC reduced thermal comfort deviation by 50.8%, electricity costs by 7.5%, grid consumption by 21.4%, and carbon emissions by 26.3% compared to standard control.
Achieved sub-millisecond inference times, recovering 87% of the MILP optimality gap at over 2,400 times lower computational cost.

Cite This Study

Ali et al. (2026) studied this question.

synapsesocial.com/papers/6a7ec707b70b84ec8b913254 https://doi.org/10.1080/19401493.2026.2712297

View Full Paper

Ask AI

Bookmark