Research on the Prediction of Traffic Accident Severity Based on BP Neural Network

Abstract

A traffic accident severity prediction study is an estimation of the expected future traffic accidents in combination with available data. One of the core objectives of this study is to predict the number and severity of possible traffic accidents so that optimal solutions can be taken in case of accidents. In this study, we propose a research method for traffic accident severity prediction based on BP neural network. Firstly, we perform principal component analysis on accident causative factors to determine the input and output factors of BP neural network; then we analyze the principle of neural network based on the composition structure, operation mechanism and learning method of BP neural network model. In the experimental part, a traffic accident dataset of Guangzhou city from 2009 to 2016 is used as an example sample for validation and the prediction results are analyzed. The results show that the prediction value of the model is accurate between 87%-91% compared with the true value, and the prediction accuracy is high, which can effectively predict the severity of traffic accidents and has certain theoretical guidance and practical significance for the prediction of the severity of future road traffic accidents.