The statistical analysis and prediction associated with nuclear meltdown accidents risk evaluation

Abstract

The relevant safety property associated with nuclear meltdown is evaluated from both reactors' internal and external factors using three statistical models: logistic regression model, linear discriminant model, and support vector machines (SVM). For each statistical model, the relevant factors that affect the nuclear reactors and probability of nuclear meltdown are evaluated by mathematical statistical analytics. Through the study, the phenomena are found that external factors have the trend to overwhelm inner factors and play a dominate role in the accident. The model analysis and their prediction results presented here could potentially provide nuclear engineers and relevant decision-makers with suggestions on selecting appropriate locations, designs and relevant construction and operation strategy for nuclear reactors from a statistical perspective.