Multi-zone HVAC control lowers peaks using edge active inference agents
One Agent per Zone: Decentralised Active Inference under Partial Observability for Multi-Zone HVAC Control on Edge Hardware
Multiagent SystemsArtificial Intelligence
Summary
Controlling heating and cooling across multiple building zones is tricky because they share equipment, so what one zone wants affects the others. The authors created independent smart agents for each zone that learn and plan on their own, without talking to each other, to keep the building comfortable while using less peak power. These agents account for system-wide effects by watching overall power use instead of direct communication. Tested on a building simulation, this approach reduced peak power and demand charges better than existing rule-based and AI methods while running efficiently on a small, low-cost device.
What this means in practice
- •For building automation engineers: Deploy independent active inference agents per zone to reduce peak HVAC power and demand charges without needing central coordination.
- •For smart building system integrators: Implement a decentralised HVAC control system capable of running efficiently on low-cost edge devices like Raspberry Pi for improved peak load management.
Authors
Panagiotis Kasnesis, Lazaros Toumanidis, Christos Chatzigeorgiou, Amalia Contiero Syropoulou, Miguel De Prado
Abstract
Multi-zone HVAC control is a shared-resource problem: zones on a floor share one fan and cooling coil, so each zone's demand affects its neighbours and the building's peak power. One controller per zone scales well and isolates faults, but each controller then observes only its own zone, not the shared air loop. In this work, we present a decentralised controller that assigns each zone an independent active-inference agent, which plans by minimising expected free energy and learns its zone dynamics online. The agents exchange no messages, so none depends on peers or a coordinator; each accounts for the unobserved coupling through a latent disturbance and a congestion signal derived from the metered power of the other floors. Fifteen agents are evaluated in EnergyPlus on a Sinergym office building against three rule-based controllers and three MADDPG fleets, one given the same floor meters. On five noisy weather years unseen during training, the fleet attains the lowest mean building peak (88.3 kW, 3 to 17 % below the baselines) and staggers the floor peaks better than any baseline, with a coincidence factor (building peak over the sum of the floor peaks) of 0.929 versus 0.936 to 0.994. Compared with a reactive rule at similar comfort (0.912 versus 0.916), it lowers the sum of monthly peaks, on which demand charges are billed, by 33 kW (4 %) in every year, at the cost of about 10 % more energy. It costs 9% less per year than the proactive rule (at lower comfort) and 1% less than the MADDPG fleet given the same floor meters, in every test year. The whole fleet runs on a single low-cost Raspberry Pi 5, closing the observation-to-action loop over MQTT in 442 ms at p95, far below the 15-minute control step.