Agder Energi and the University of Agder have achieved concrete advances in simulations for artificial intelligence controlling hydroelectric power production. Minister of Research and Higher Education Iselin Nybø was thrilled about what she saw and heard.
This article is more than one years old, and may contain outdated information.
The primary goal of Agder Energi and the University of Agder’s research project is for artificial intelligence to control hydroelectric power production better than a human could. Ole-Christoffer Granmo, researcher and Director of the Centre for Artificial Intelligence Research (CAIR) at the University of Agder, says there is a lot of uncertainty in today’s electricity production.
"It is difficult to plan water supply and price of electricity, and therefore it is also challenging to find to most optimal use of hydroelectric power. When artificial intelligence can control more of the production, the idea is that it can control a hydroelectric power plant better than a human could. Recently, we have made concrete advances that show that this could become reality in a few years," Granmo says.
Iselin Nybø, the Minister of Research and Higher Education, visited the power plant at Hunsfos in Vennesla, and she was thrilled about what saw and learned.
"This is a very good example of how academia and the industry should collaborate. It is very exciting and future-oriented that academia and Agder Energi come together to optimise the use of water for hydroelectric power," Nybø says.
Researchers are simulating billions of scenarios to develop control systems that can decide when water should be used for producing electricity. The mathematic models that currently control electricity production are about 30 years old, and optimising the production to be as lucrative as possible is challenging. The goal is that artificial intelligence will contribute to production being as self-learning and optimal as possible.
"The University of Agder is making significant investments in our expert environment, and we are now investing in nine super computers to increase our effort in developing artificial intelligence. The EU is also investing large amounts of research funds in this field of expertise, and getting access to a part of these funds is something we aim for, of course. If we succeed, we can get a technology that we, naturally, will be able to export," Granmo says.
Machine learning is one of the areas the research centre makes an effort, and it is an important part of the collaboration project with Agder Energi. With help from so-called deep reinforcement learning, the computer can on its own develop methods and models that take inaccuracies and uncertainties considering the water supply into account.
"We always want to improve our electricity production and the way we use our water resources. The University of Agder is at the front in Norway in its work on artificial intelligence. Therefore, we consider the collaboration to be very exciting. There are also great possibilities here for us for higher profits," Bernt Viggo Matheussen, head of Agder Energi’s division of hydrology and meteorology.
Matheussen has during the project’s duration held a part-time position at the CAIR research centre and works closely with Granmo in developing the models. The research project Generic Hydropower Optimization Using Deep Reinforcement Learning is supported by the Research Council of Norway’s ENERGIX programme.
"The development in this area goes incredibly fast, and it is a race to stay in the lead. But if we can succeed in getting our computer programs to control the process on their own, we will be the first to implement this technology for controlling the production of hydroelectric power," Matheussen says.
The CAIR research centre was opened in March 2017 at the University of Agder. Today, there are 14 researchers with varied backgrounds connected to the centre. At the same time, artificial intelligence is a field of expertise that has been built up over a long period of time at UiA. However, in recent years, the development in this field has progressed at a record pace.
"Currently, there is a revolution within research on this field, and artificial intelligence is increasingly becoming an important factor for the social development. We will continue to be at the front in this field, so we must always continue to develop our research and our methods," Ole-Christoffer Granmo says.
The three priority areas are machine leaning, reasoning and natural language. CAIR also has research projects in collaboration with the Hospital of Southern Norway, and it is a partner in a larger project managed by SINTEF regarding developing computers that can provide advice on mental health, so-called chatbots.