Traffic Accident Prediction based on CNN Model

The traffic accident prediction represents a vital role in the combined planning and management of traffic, the design which involve some nonlinear components, such as people, road, vehicle, weather and so on. The convention way of extended studies cannot expose the real situation since the noise corruption and amount of data are too small, so that the result of prediction cannot be satisfied. This paper proposes deep learning-based new road traffic accident prediction applying a “Convolutional Neural Network model” (CNN). It uses traffic accident influencing circumstances like light, weather, traffic flow to make a state matrix describing the traffic state and CNN model. This paper utilizes examples to examine the accuracy of the proposed model. The empirical results show that the proposed model is more efficient than the current neural network design to predict traffic accidents compared with the traditional “Backpropagation” (BP) algorithm