MURMANSK, October 23. /TASS/. Scientists at the Murmansk Arctic University (MAU) developed an energy consumption forecasting model that may improve reliability of energy supply in the Arctic's conditions of high social and technical vulnerability, the university's press service told TASS.

"MAU's Department of Energy and Transport Construction has been implementing a unique project to forecast energy consumption, which is especially important due to the expanding infrastructure and industrial activities in the Arctic," the press service said. "The university's scientists have developed a hybrid forecasting method that combines decision trees, regression analysis and adaptive boosting coefficients for every node of a tree."

According to the project leader, senior lecturer Nikolay Nagibin, modern energy consumption forecasting models are unable to cope with the Arctic conditions, like, for example, sharp climatic fluctuations, low population density and high cost of errors. "In the Murmansk Region, an underestimated forecast may lead to emergency shutdowns in extreme frosts, while overestimation causes only minor energy losses, hence the asymmetric cost of errors. Existing approaches do not envisage this asymmetry, which makes them unsuitable for the Arctic energy," the scientist said.

The new model, developed at the Murmansk University, purposefully minimizes the risk of forecast underestimating, while maintaining a moderate overestimation. This approach is based on a statistically verified correction of forecasts, taking into account local consumption specifics, thus ensuring reliable energy supply. The proposed method can become the basis for sustainable and safe energy management in the Arctic's remote areas, the university's press service said in conclusion.

The Murmansk Arctic University, created as a merger of two major universities in the Murmansk Region, has received more than 1 billion rubles ($12 million) for creation of a Coastal Training Center and the upgrade of the Marine Convention Training Center.