AI for home energy management

How simple the old days were: in the spring, when heating oil prices dropped, you simply filled the tanks in the basement to the brim. Then you were free of all worries until the next season. There was also fuel for the car on every corner. Around the clock. Fill it up, drive on, and you're done.

The phase-out of fossil fuels makes it much harder for people who want to save money. Energy prices no longer change annually, but hourly. Solar power is plentiful at lunchtime - in the evening, the low sun hardly supplies any energy, and at the same time, commuters returning home are causing electricity demand in cities and rural areas to rise rapidly. The effect is so clearly visible on consumption graphs that scientists have given it its own name: the "duck curve". When the duck rears its head, it becomes expensive for everyone who now has to purchase electricity.

Watching the clock when drawing energy would therefore be important for electric car drivers and homeowners. In the future, anyone who wants to use the available renewable energy in a cost-effective and at the same time environmentally friendly way will no longer be able to rely on permanently installed thermostats and manually operated buttons.

A multifaceted problem

Bratislav Svetozarevic, a researcher in the Urban Energy Systems Laboratory at Empa, has recognized the problem. What is needed is an automatic control system that stores energy at favorable times of day and makes it available for expensive times of day. For example, the drive battery of one's own car, which is connected to the charging station in the garage, could serve as a storage device. But Svetozarevic is dealing with a multi-faceted problem: Every house is different, and so are its occupants. Moreover, depending on the weather and the season, the electricity generated by the solar panels changes, as does the demand for heating or cooling power. An optimal energy control system must therefore learn the daily rhythm of a house and its occupants - and should also be able to react flexibly during operation, for example if a change in the weather overturns all calculations.

Step one: the theory

The solution to such problems is artificial intelligence. The Empa researcher designed an AI control system based on the reinforcement learning principle. When the system acts "correctly", it receives a "reward". Gradually, the controller perfects its behavior in this way.

Initially, the control system was only simulated on the computer. The specifications: A specific room in a building had to be heated electrically to the desired temperature and maintain it. At the same time, the system had to supply electricity to an electric car, which was to be at least 60 percent charged at 7:00 in the morning and set off on its journey. In the evening at 5:00 p.m., the electric car returns to the charging station with a remaining charge and can also deliver power back to the house during the night hours. The control system was fed with weather data and room temperatures from the previous year and had to cope with two electricity tariffs: expensive electricity during the day between 8 a.m. and 8 p.m., and cheap electricity during the night hours.

The result was astounding: the self-learning control system saved around 16 percent energy compared to a fixed-program solution and also maintained the desired room temperature much more precisely in the theoretical test.

Step two: Test in real building

Now the control system had to pass the test in reality. Svetozarevic used NEST on the Empa campus for this purpose. In the DFAB House unit, the AI algorithm controlled the temperature of a room for a week. At the same time, the 100 kWh-sized storage battery in NEST was used to simulate the battery of the electric car. This time, the result was even more pronounced: In a cool week in February 2020, the AI control saved 27 percent heating energy compared to the neighboring student room, whose heating was operated with a fixed-programmed (rule-based) control.

"The beauty of our self-learning AI control system is that you can use it not only in the NEST research building, but also in any other building," says Bratislav Svetozarevic. "It doesn't need an engineer to program the control system, and it doesn't need anyone to analyze the building beforehand and calculate a customized solution."

Comfortable warmth in an economical way 

In a next step, Svetozarevic and his colleagues now want to determine how the system can be extended from one room to larger buildings. "In our first experiment, we wanted to model a typical household of the future," says the Empa researcher. For simplicity's sake, the team limited itself to heating and vehicle charging. However, the work lays the groundwork for much more. Svetozarevic is certain: "Our AI controller will still be able to cope when a photovoltaic system supplies electricity, a heat pump and a local hot water tank have to be operated - and the occupants' comfort requirements keep changing."

However, in order to be able to use the AI system for an optimal energy supply in the future, a new generation of electric cars is needed. Today's standard European and U.S. models with the CCS fast-charging connector can only fill up with electricity, but cannot supply it. Japanese cars with Chademo plugs, on the other hand, are designed for bidirectional charging. In December, the Korean company Hyundai announced that its new E-GMP electric car platform would also be equipped for bidirectional charging. This could help electric cars save energy in the long term and at the same time stabilize the electricity grid.

For more information: www.empa.ch/web/energy-hub