Prediction Model for Ship Collision Accidents in Port Waters Based on Principal Component Analysis and Convolution Neural Networks

In order to improve the prediction accuracy of ship collision accidents in the waters of Tianjin Port, this article constructs a PCA-CNN ship collision accident prediction model based on principal component analysis (PCA) and convolution neural network (CNN) principles. Based on historical collision accident data in the waters of Tianjin Port, key factors affecting ship collisions were screened using PCA. On this basis, CNN models were used to predict collision accident and non collision accident data, and compared with other prediction models. The results showed that the proposed model had the best predictive performance, with an accuracy of 85.71%; the prediction efficiency is the highest, with a training time of 8.64s