On Automatic Identification Approach For Road Alignment

Digital geography traffic information service was required of rendering network map, positing vehicles and intelligent traffic system which are realtime update, accurate and detailed because of the increasing growth of the vehicles. In the recent years, the country enlarged the risk of the based structure. The speed of the roads' newly building, widening and rebuilding becomes quickly and quickly. The existing road information which is inaccurate and incomplete, that is bad up-to-date state becomes the problems that need to solve.The paper was supported under the Natural Science Foundation of China "A identification method of automobilism tendencies based on collaborative deduction of man-vehicle-environment dynamic data", the Natural Science Foundation of Shandong Province "A inference approach for road safety based on traffic situation" and the Science and Technology Development Program of Zibo "A threat assessment method for urban road". Road information was studied in particular include classifying road alignment automatically and personalized travel information for traveler by different algorithms.Updating road information database in the sustainability, methods about road alignment automatic classification based on the Incremental Bayes Classifier, LVQ-Boosting model and modified SLFNs algorithm were proposed, which utilized a large number of path tracking trajectory data generated by car GPS for capturing quickly changes in road information.The Incremental Bayes based on data stream classification and incremental learning was applied into bayes algorithm. LVQ-Boosting algorithm was improved the generalization ability of LVQ and obtained a classifier with strong classification performance through employing weak classification algorithm. Besides, an efficient hybrid classifier was built, which integrated the one-pass clustering and the single hidden layer feedforward neural networks (SLFNs), and proposed a training algorithm uses singular value decomposition to calculate the network parameters. It is simple and has low computational complexity. In addition, the paper was also analyzed the impact of variability in road traffic on the road type of extraction by means of road traffic conditions and environmental changes, and investigated an approach to quickly and accurately capture road changes, to classify the feature of road type and to extract the new road.The three approaches were taken the GPS positioning coordinates, the speed, the radius of horizontal curvature and the travel direction as the basic identification features and input variables. The features of the road alignment include straight road, curve, T-Road, intersection and roundabout were taken as output variables. Thus, the purposes for identifying automatically the features of the road alignment and fast grouping the road feature type were implemented. The experimental results showed that the three approaches has high efficiency and accuracy of the road alignment identification and the identification capacity of the modified SLFNs algorithm is the best.Futhermore, in view of traveler differences in individual needs and a case of a traveler, the travel information of travelers was quantitatively analyzed and their various individual travel requirements were summarized through mining laws of traveling information and generating the classification decision tree and the association rules with Weka analysis platform. Thus, the classification results of travel information were obtained by classification analysis, the correlation between individual travel information was obtained by correlation analysis, and the relative aggregation of travel information was presented by cluster analysis. While traveler faces massive information, it can remit the extreme reflection, gathering reflection and so on. It is adopted to obtain traveler's personal request of travel information. Thus, the travelers can be offered the information service which is convenient and intelligent.