Norway works on AI project to control hydropower plants

Norwegian scientists are working on a ground-breaking artificial intelligence (AI) project on automatic control of hydroelectric power plants, Norway's leading newspaper Aftenposten reported Sunday.

"The major challenge is to produce power when there is high price to release water until there is a danger of flooding," said Ole-Christoffer Granmo, professor and director of the Centre for Artificial Intelligence Research (CAIR) at the University of Agder, which works on the computer program that will control water levels in hydroelectric plants.

"When the computer program which we develop gets an overview of weather data, prognosis, weather forecasts and water level in real time, it will be able to control ponds at the power plant without any human intervention," Granmo said.

If the scientists achieve their goal, they will be first in the world with an artificial intelligence that controls the hydropower plants, Aftenposten wrote.

According to the newspaper, students at the Norwegian University of Science and Technology (NTNU) in Trondheim are also working on two ground-breaking artificial intelligence projects -- on monitoring of electrical grids without intervention of people and automatic navigation of submarines.

The project on electrical grids aims to replace humans with AI to detect errors occurring in the grids by analyzing images taken from the air.

"By automatically analyzing of the pictures, you will get the same results every time, while different people will make different assessments. It can be compared with the various Norwegian teachers taking the same Norwegian language style in a different way. The result is that machines alone can detect errors in the grid much faster and more accurate than today," said Per Magnus Veierland, a master's student at the NTNU who's working on the project.

Veierland's fellow student, Hakon Gimse, has been working at the NTNU on artificial intelligence that can be used in submarine navigation.

A submarine must always have an overview of other vessels in the vicinity in order to avoid collision. Today this is done manually by a sonar operator aboard the submarine, listening to the sounds of the vessels around.

"The task is difficult even for a trained operator, because there is often a lot of noise in the water from whales, fish and other vessels," Gimse said.

He has developed a system based on several types of deep neural networks that are able to learn this task.

"I have done it by showing the network several thousand examples of how different vessel sounds, to form a general understanding of how a type of vessel should sound like. The result is a program that can be used to listen in all directions simultaneously and estimate which vessels surround the submarine in a both faster and more accurate way than humans are capable of," Gimse said.

He has developed a prototype, which can also be used in other contexts, for example, to say something about what kind of vessels are going in and out of a harbour area. Endit