- System uses meteorological data and artificial intelligence to predict the occurrence of lightening
- The data that is gathered for prediction comprises of atmospheric pressure, air temperature, humidity and wind speed
- The machine can predict lightening half an hour before and within periphery of almost 36 kilometers
According to a survey, lightening is considered as the most unpredictable phenomenon in nature until now.
By merging meteorological data and artificial intelligence, students have determined the time and place of the occurrence of lightening. The prediction will be done 30 minutes prior to its occurrence. The machine has 80% accuracy.
The technology is set as a warning system to prevent the effects of lightening strikes to critical infrastructure, sensitive equipments and outdoor facilities.
Amirhossein Mostajabi, a Ph.D. student who had formulated the technique said,” Current systems are slow, complex and require expensive external data acquired by radar’s satellite.”
“Our technique utilizes the data can be obtained from any weather station. So, with the help of this technology, we can cover remote regions that are not within the radar along with the places that possess communication breach.
The researchers’ have used the methods of machine learning algorithm that has been accustomed to recognize conditions that lead to lightening. For prediction, it analyzes atmospheric pressure, relative humidity, and wind speed.
The dataset used in the ML model comprises of data utilized as predictors, meteorological data to name a few. The data from the lightening system are used to train the Machine Learning model and then validation of accuracy of lightening takes place to show the competitive base lines.
The method is an approach that helps in forecasting the most unpredictable phenomenon-lightening.
Unlike other systems that are based on data obtained networks used to detect lightening, this model gives a tool for predicting the occurrence without using previous lightening Data as a precursor of the warning.