Integer Model Predictive Control
AI-driven predictive control based on physical models of your building
Resolve network congestion without costly upgrades
Up to 40% savings on energy
10 days of model training time
Use of total electrical power.
Total power with conventional control:
Demand for heating and cooling peaks during the day, causing total energy consumption to exceed limits when other electricity consumption also peaks at the same time.
When using Integer MPC:
Our model combines controllable (HVAC and possibly batteries) and non-controllable power consumers and generators (solar panels, lighting, office equipment) to ensure that the grid connection is not exceeded, while optimizing for a configurable cost function—such as energy costs, CO2 emissions, or comfort.
Indoor temperature control.
Temperature control with conventional control:
Overheating in the early morning is caused by the heat pump’s delayed response to outside temperatures. This is exacerbated by increasing internal heat load and solar radiation. Cooling is activated to counteract this. The slow response continues and perpetuates the cycle.
Temperature control with Integer MPC:
Our predictive model anticipates the required heat demand and keeps the indoor temperature just above the lower limit to optimize energy consumption. Solar gain increases the indoor temperature, but cooling is not necessary because the temperature remains within comfort limits. The residual heat from the solar gain dissipates slowly, which significantly reduces energy consumption.
Preventing peaks in heat demand.
Peaks in heat demand with traditional control:
Heat demand increases when the temperature drops significantly (at night), particularly in the morning. Peak consumption occurs in the early morning, at the same time as other electricity consumption, causing grid congestion. As the temperature rises during the day, heat demand gradually decreases.
Solving network congestion using Integer MPC:
Our predictive model anticipates limits due to grid congestion or the maximum capacity of the heat pump and starts heating much earlier to spread demand. The system reduces consumption during peak times and increases capacity again when surplus solar energy becomes available later in the day. Heating capacity then gradually decreases throughout the day, as in the traditional case.
Implement smart predictive control within days, not months
We combine machine learning techniques with physical modeling to significantly reduce implementation time.
Is the challenge you are facing not mentioned above?
We can customize our control algorithms to help you with your specific challenge.
Let's have a chat
