벡터자기회귀(VAR)을 이용한 감마선량 이상상황 예측

초록

Gamma rays are very dangerous to the human body and are contained in natural or artificial substances. The gamma dose can be continuously measured to determine whether there is an abnormality in the release of radioactive material in a certain space. Some weather-related information is correlated with gamma doses in space, and the dataset used to predict gamma doses requires that a multivariate data model be constructed with gamma ray measurements and multivariate amounts of weather data. The Vector Auto Regression (VAR) model extends the univariate auto regression model to multivariate time series data, making it one of the most widely used probabilistic process models for analyzing the interdependence of multivariate time series, and is useful for explaining and predicting the behavior of time series. In this paper, we use the public data gamma dose and weather data, and confirmed that it is possible to predict anomalies of spatial gamma doses according to weather changes through the VAR multivariate data model.

키워드