Occupancy-driven predictive control of HVAC systems

In institutional buildings, where occupancy profiles are highly variable, HVAC control must reconcile energy reduction with thermal comfort. Most existing approaches rely on deterministic, supervised occupancy models that poorly capture uncertainty.

This doctoral project proposes a probabilistic occupancy-driven approach: an unsupervised, adaptive model based on hidden semi-Markov models (HSMM), able to represent occupancy dynamics and state durations, integrated into a stochastic model predictive control (SMPC) framework. The goal is to explicitly exploit occupancy uncertainty to optimize control decisions, reduce energy consumption and preserve thermal comfort.

Main objective

Propose a stochastic model predictive control (SMPC) strategy for HVAC systems that guarantees thermal comfort while accounting for occupancy and weather uncertainty in institutional buildings.

Methodology

Three stages: review of occupancy estimation methods; development and validation of an unsupervised occupancy model based on hidden semi-Markov models (HSMM) from sensor data; integration into an SMPC framework evaluated by simulation.