A Traffic Accident Predictive Model Based on Neural Networks Algorithm and Rough Set Theory

The most important and critical step to improve road traffic safety is prediction and identification of traffic accident black spot. A new prediction model of traffic accident black spots is proposed based on GA-BP neural network algorithm and rough set theory. First of all, the traffic accident statistics of Jinwei Road in Tianjin are analyzed. With consideration of static road conditions, the samples of road accident black spots are obtained by the GA-BP neural network algorithm. Furthermore, an effective road traffic accident black spot prediction model is established by utilizing rough set theory with consideration of the impact of real time dynamic conditions. Finally, a numerical example is illustrated. Experimental results show that the proposed model with the combination of these two theories can reduce the hybrid and burdensome amount of data, lower the false alarm rate and improve the forecasting accuracy of accident black spots.