The Application Research Of Fuzzy Clustering And The Traffic Black-Spot Cause Of Formation Analysis

Among all kinds of accidents, traffic accidents are always on the top rank, not only on frequency but also on casualty, and it has a tendency of climbing. The casualty of traffic accidents takes about 70 percent of the total accidents. Since the casualty of traffic accidents broke the amount of 100,000(160thousand) in 2001, traffic safety has become an urgent problem which should be solved in transportation realm. As a high frequency section of traffic accident, traffic black-spot cause tremendous economic loss and casualty, which take a great part of the total. The reason for black-spot forming is not in one aspect. It always caused by two or more reasons. How to confirm black-spot, how to analyze black-spot and how to check them are difficult points for present traffic management. Researching and analyzing reasons of black-spot forming and its influence can not only provide decision-making supports for accident black-spot preventing, but also have great meaning for reducing traffic accidents frequency, enhancing traffic management effectiveness and promoting general impression of our country's traffic.As an important section of data mining, clustering is an effective tool for analyzing large amount and complicate data and a major means for get intelligent decision-making information. The traditional means of getting information is mining knowledge from single data set, which neglects the inter-influences of multi-data sets. However, the reasons for accident forming black-spot are not single, which have lots of independent but inter-influencing factors. As the traffic accidents are more and more serious and accident black-spot are increasing, using information cooperation for reason analyzing, which can indicate more clear of real situation seems to be more and more important. On the basis of fuzzy clustering analysis, this paper takes black-spot of an accident as example and researches deeply the application of fuzzy clustering that based on information cooperation in traffic black-spot forming. The research performances are as follows:1. Various initial algorithms of fuzzy clustering are analyzed and compared, and then an improved initial algorithm is put forward to optimize the fuzzy clustering number and the initial clustering center by computing the average information entropy and combining with subtractive clustering.2. The information cooperative clustering model has been chosen to do analysis between data sets cooperately, an improved cooperative fuzzy clustering algorithm is put forward with improvement on the selection method of the initial center and clustering validity function aiming at applications, the experiment result indicates that the improved cooperative fuzzy clustering algorithm precede the original algorithm on iterative times and much more approaching to the class center.3. A method on traffic black-spot cause of formation analysis based on cooperative fuzzy c-means clustering is proposed, analyzed and then realized, an example analyzing is made using a certain case of a traffic black-spot, the result indicates that the cooperative analyzing method can be used to assistant decision-making in practice.Finally, the work of the full text and creativity are summed up, and the expectation towards future research is given at the end.