Identification for traffic accident severity at urban road entrance within complex road network environment.

In the urban road entrance accident severity identification process, data processing inadequacies often lead to a high rate of false and missed recognitions. To address this issue, an innovative method for identifying traffic accident severity at urban road entrances within complex road network environments has been developed. The method involves smoothing and enriching multi-channel collected urban traffic data, followed by optimization through an empirical cumulative distribution function. A complex road network model is then constructed to monitor the flow at the entrance, integrating a Hybrid finite mixed model enhanced by the Expectation-Maximization (EM) algorithm to achieve accurate accident severity identification. Experimental results show that the maximum misidentification rate for accident severity at urban road entrances is 0.7%, with a maximum missed recognition rate of 0.6%, and an average delay time of 0.77 seconds, and confirm the reliability of the identification results. [ABSTRACT FROM AUTHOR]