Experts from Lausanne, developed the technology of machine learning, which will enable artificial intelligence systems to predict the time and place of a lightning strike 30 minutes before the occurrence of natural phenomena.
The system will rely only on the weather.
Scientists noticed that all the existing systems predict such phenomena are slow.
Furthermore, their work implies the need of a specialized source of data received from radars or satellites.
Experts came to the conclusion that lightning occurs in specific conditions that can be predicted and calculated through the application of an artificial intelligence system.
The new technology, according to the developers, will operate with data received in real time from any weather station.
Thus, specialists will be ready to give an accurate prediction for the most distant geographical points of the planet.
This technology has a lower accuracy, but greater availability and scope of the prediction of lightning strikes.
Forecasts system is based on the study of cloud types and meteorological conditions that lead to such phenomena.
Physicists picked up a number of machine learning algorithms.
The data was based on gradient boosting technique.
This method combines chain data from multiple systems, inaccurate predictions that can learn on erroneous results and gradually to give the correct answer.
For machine learning technology had prepared a special dataset, which for several decades has been collected by the workers of meteorological stations.
Measure artificial intelligence system able to recognize the characteristic features of the phenomenon and had the ability to predict the zipper on the change in air temperature, humidity and pressure and other meteochannel.
Testing machine learning systems has allowed to reveal the accuracy of the predictions of development.
It turned out that in 76 % of cases, the system predicted lightning strikes 30 minutes before their physical appearance in the area of weather stations.
The application of this technology can significantly improve the safety of air travel, to ensure the efficient operation of electricity grids and other infrastructure.