Abstract:
Based on the traffic accident data and meteorological observation data around the Hang-Jinqu expressway from 2006 to 2016, the temporal distribution of traffic accidents on the Hang-Jinqu expressway and its response relationship with meteorological conditions were analyzed. Finally, the traffic accident index meteorological forecasting model was established, taking its autocorrelation into consideration. The main conclusions are as follows. (1) The number of traffic accident happened on rainy days are the highest among the all five adverse weather conditions on the Hang-Jinqu expressway, and snow weather conditions most easily leads to the occurrence of traffic accidents. (2) In winter, meteorological factors have the greatest impact on traffic accidents, and temperature, relative humidity, precipitation, wind speed, and visibility are all significantly correlated with the traffic accident index. Meanwhile, meteorological factors have the weakest impact in autumn when only precipitation significantly correlates to traffic accident index. (3) On the basis of different meteorological factors in different seasons, accident index forecasting models are established. It is found that accident index observation and forecasting values fit the best in winter, and the model has a good performance in moderate, sub-high and high accident index levels. The prediction accuracy of moderate and high accident index levels are 66.1% and 50.3%, respectively. Thus, the model has good application in traffic weather service.