Research On Traffic Flow Of Extra Long Highway Tunnel

For traffic flow of extra long highway tunnel contains large traffic information, it is necessary to find the useful information by data analysis on the traffic flow of extra long highway tunnel. Clustering on traffic flow of extra long highway tunnel is expanded by the improved fuzzy c-mean clustering algorithms. The abnormal data caused by the failure of techniques or equipment in the data collection or transmission have reduced the quality and credibility of data. Research on the abnormal data of traffic flow of extra long highway tunnel becomes necessary, and it is important to increase the quality and credibility of traffic flow data of extra long highway tunnel by detecting and amending the abnormal data.Considering the characteristics of nonlinear,complexity and time-variables of traffic flow, short-term prediction on traffic flow of extra long highway tunnel is well researched in order to improve the prediction results. Along with the unceasing consummation and development of chaos theory, it is capability for research the chaotic characteristics of traffic flow of extra long highway tunnel.After data analysis on traffic flow of extra long highway tunnel, the expended research on traffic flow of extra long highway tunnel is focused on the four aspects:fuzzy clustering, abnormal data, short-term prediction and chaotic characteristic of traffic flow. The new points of this dissertation are as follows:(1) Research on the clustering on traffic flow of extra long highway tunnel, an improved fuzzy clustering algorithm is put forward. The objective function of the improved clustering algorithm is optimized by the amendment of membership function and distance measuring function, which can weaken the impact of noise on the clustering result. The automatic clustering is obtained by the degree of cohesion and separation. To apply the improved fuzzy clustering algorithm, the fuzzy clustering and the clustering center of the traffic flow of extra long highway tunnel are obtained.(2) Research on the abnormal data of traffic flow of extra long highway tunnel, based on the characteristic of the traffic flow indictors including traffic volume, traffic speed and time occupancy, a new method of abnormal data detection on traffic flow of extra long highway tunnel is presented. The historical data on traffic flow are clustered according to the average xaffic speed by the improved fuzzy clustering, then the algorithm on the determination safety zone scope is obtained and abnormal data can be detected from the real-time traffic flow data. And the abnormal data can be eliminated or amended using the improved fuzzy clustering algorithms, which can enhance the quality and reliability of the traffic flow data of extra long highway tunnel.(3) Research on the short-term prediction on traffic flow of extra long highway tunnel, an improved recurrent compensatory Neuro-fuzzy system is established based on both the gradient descent method and the weighted recursive least squares estimation method. To complete the auto update clustering using the improved fuzzy clustering algorithms; to introduce the recursive weight coefficient for the time-delay characteristic of traffic flow; to add the compensation coefficient for automatically correcting the membership; to apply the gradient descent method and the weighted recursive least squares estimation method for parameter learning; to take the clustering center as the initial input center of the fuzzy neural network. Using the actural traffic flow data of extra long highway tunnel, the system established in this dissertation improves the prediction efficiency and provides the more timely and effective future traffic flow information for the research on traffic safety of highway tunnel.(4) Research on the chaotic characteristic of traffic flow of extra long highway tunnel, existence of chaos in traffic flow of extra long highway tunnel is proved by the improved identifying chaos method based on a small quantity of data. The chaos in traffic flow of extra long highway tunnel is applied to the improved recurrent compensatory Neuro-fuzzy system based on Hybrid Learning method, and then the efficiency of the system is improved. Through the research on the chaos in traffic flow of extra long highway tunnel, it verifies not only the existence of chaos in traffic flow of extra long highway tunnel but also the active role of chaos in research on traffic safety of extra long highway tunnel.