Helen and Silo AI bring intelligence to district heat production in Helsinki

Helen, the electricity and heat energy producer in Helsinki, and Silo AI have together developed an AI-driven solution that improves forecasting of district heat demand and makes production planning more efficient. The intelligent solution learns from historical data and is able to make demand forecasts up to 40 hours. The results of the solution, already implemented in Helsinki, are promising as it reduces error rate between heat forecasts and actual demand by one third.

Most of the houses in Helsinki are heated with district heating all year round. With the machine-learning based solution Helen can now better plan its heat production throughout the year.

The intelligent solution leverages historical forecasting data in addition to weather and time related data. The historical data covers both previous demand forecasts made and actual demand for the last two weeks. In addition, the solution adapts to new data and improves over time.

The intelligent application enables the planning of district heating production as close as possible to the actual demand. This is important in order to produce the right amount of district heat.

– The solution has been piloted in Helen alongside the current forecasting system, and the results are promising. The quality of the heat demand forecast will further improve over time as the data increases. By utilizing the latest technologies, Helen brings the opportunities of a new era in energy to everyone. As the entire energy system becomes more efficient, our customers also benefit. Helen's goal is carbon-neutral energy production, and the efficiency of the entire energy system contributes to reducing carbon dioxide emissions, says Janne Inkinen, Advisor at Helen.

– It is great to see that Helen is boldly developing new technology for a challenging problem  in collaboration with us. Creating foresight based on dynamic data is a task where machine learning significantly benefits both heat producers and consumers, says Peter Sarlin, CEO of Silo AI.