Risk assessment method for dangerous goods transportation by sea based on generalized regression neural network

This paper discusses the application of a generalized regression neural network (GRNN) in the risk assessment of dangerous goods transportation by sea. As a type of neural network that is based on a radial basis function, the GRNN is suitable for processing small sample data with nonlinear characteristics. In this study, a sample dataset containing seven risk factors was constructed, and the SPREAD values were adjusted to optimize the performance of the model. The experimental results show that when the SPREAD value is 0.3, the GRNN model has high prediction accuracy and good generalizability for both the training and test samples. Therefore, the GRNN model is applicable and reliable for assessing the risk level of dangerous goods accidents, which provides a new assessment tool for risk management of dangerous goods transportation by sea.