In order to better analyze the influencing factors of traffic jams in some sections and optimize the accuracy of traffic jam prediction, we use the collected data to predict and analyze the inherent regularity of traffic jam duration. We establish a model for different situations and based on the data of the query, using principal component analysis, mathematical statistics, queuing theory, operations research and other knowledge to give certain algorithms, and draw conclusions. First, this article establishes a principal component analysis model. the model mainly includes the traffic factor and the time factor. Using this analysis model, we can calculate the linear combination of all the two components as the individual variables. We concluded that the average speed has the greatest influence on the degree of traffic jam, and the influence degree is as high as 94.1%. In the time factor, the influence of the early peak on it slightly higher than the evening rush hour, the proportion is as high as 84%. Secondly, based on the main influence factors of the model one, we construct the queuing model of the vehicle on the congested road section. we decompose and analyze the composition of the congested road queuing system, optimize the road queuing system, and increase the influence of the jamming behavior to predict the vehicle when it is crowded. Waiting time to ensure the right service intensity, and is supported by specific analysis. According to the case analysis of the Yagang South Avenue in Guangzhou, the four indicators in the original model have changed: because the occurrence of the jamming behavior, there is a decrease in the value corresponding to the original model of 0.09 and an increase of the value of 0.01, the average waiting time of the whole process is the time when the vehicle was waiting in line to reach the first row and the time passed the green light rose by 1.3 minutes; the average waiting time for the queue rose by 0.9 minutes. Finally, the paper analyzes the error of the model, and obtains the comp rehensive evaluation model in the sensitivity analysis. This article also introduces the promotion and evaluation of the model.