Featured Project
Supervised Machine Learning Energy Consumption
Supervised Machine Learning Energy Consumption is a proof of concept of an AI-driven forecasting system to help save energy, cut costs, and improve comfort. Starting with Long Wing 1 of NATO Headquarters, it is designed to scale across NATO facilities and extend into prescriptive models for automated optimization.
Background
Today, the HVAC system (heating, cooling, ventilation) operates in a reactive way, it responds only after conditions change. This is costly, less efficient, and sometimes overcompensates, wasting energy and straining equipment.
ACT Innovation built a predictive intelligence layer on top of the existing system. Using building sensor data, weather forecasts, and occupancy patterns, we trained AI models to forecast comfort (temperature & humidity) and energy needs ahead of time.
What We Delivered
Predictive Models
AI models that forecast future temperature and humidity across the building.
Enthalpy-Based Savings Screening
Hour-by-hour cooling/heating opportunity estimates using moist-air enthalpy derived from the T/RH forecasts, to size potential savings before any control changes.
Scenario Engine Outputs
Side-by-side results for Baseline, Max-Optimum (theoretical), Smooth (feasible), and Scenarios (incremental setpoint tweaks) with comfort and enthalpy-proxy KPIs.
Comparative Analysis Report
Side-by-side comparison of current vs. predicted optimized scenarios (comfort + enthalpy-proxy) in 2024.
Validated Dataset & Documentation
Cleaned datasets, data dictionary, and model/feature documentation for reproducibility.
Project Execution Report
Methods, assumptions, results, and next-step recommendations (including the pathway to future calibrated energy modeling).
Main Benefits
Energy Efficiency
Reduce energy usage by avoiding excessive system responses. More stable HVAC performance with fewer peaks in consumption.
Cost Savings
Lower operational costs through smarter system behavior.
Minimize overcorrection due to delayed thermal responses.
Improved Comfort
Maintain indoor temperature and humidity within comfort zone.
Avoid discomfort caused by overcooling or overheating.
Scalability
What we prove in Long Wing 1 can be rolled out across other HQ buildings and NATO facilities, creating a model for energy-smart military and administrative infrastructure.
Outcome
The optimizations the model brings are expected to have a financial and sustainability impact in the Operation and Maintenance of the NATO Headquarters, as well as improving the occupant well-being.
Additionally, the methodologies and technologies employed for this project can serve as a foundation for similar energy optimization projects across other NATO buildings.