Cloud-Based Real-Time Model Predictive Control for a Multi-Carrier and Multi-Objective Home Energy Management System

Integrating thermal and electrical demands in residential energy systems is challenging due to dynamic usage patterns and resource constraints. This paper introduces a cloud-based home energy management system (CBHEMS) leveraging real-time (RT) optimization and Model Predictive Control (MPC) to coordinate multi-carrier micro-energy hubs (mEH). The proposed approach enables bidirectional exchange with the electricity and district heating networks, integrating RT load and photovoltaic forecasts with dynamic pricing. An Apache Kafka-based communication infrastructure and InfluxDB manage data flow and storage, while containerized forecasting and pricing services run on Kubernetes (K8s) for scalability and reliability. Using a receding-horizon MPC framework, the CBHEMS solves a linear multi-objective optimization problem via the augmented epsilon-constraint method (AUGMECON) to minimize operational costs and carbon emissions. The simulation results demonstrate that the system adapts effectively to varying conditions, achieving higher energy efficiency, cost savings, and reduced RT emissions. A sensitivity analysis also highlights the impact of prediction horizon length on performance. Comparisons with traditional day-ahead scheduling highlight MPC’s robustness in handling uncertainties and leveraging continually updated data. This work marks a significant advance in the deployment of intelligent, sustainable, and scalable cloud-based energy management solutions for modern residential environments.