The Data Mining Of Urban Road Traffic Flow

Urban traffic flow which shows the characteristics and patterns of representation through a large number of traffic flow information comprehensively reflects the situation of urban transport. It is the essential scientific basis for traffic management and decision-making. It is also the main travel reference for urban people. As the traffic flow contains a large amount of information that can be used for improving traffic operating conditions, so how to mine its potential value and the law is one of the hot researches. In this paper, we analyze traffic flow information in two aspects.(1)The similarity analyze of traffic flow. The previous studies of traffic flow are almost based on the time series, such as:the research based on weekly traffic flow. In fact, traffic flow information is space-time correlation. So we need to consider both time and space factors when analyzing traffic flow information. The traffic flow between the crossings has the following relations, such as:the similarity of adjacent crossings is maximal, and is much larger than others; similarity is inversely proportional to distance, that is to say the longer distance, the smaller similarity etc. With this useful information, traffic managers can make a better management. As distance is the vital factor, so in this paper we take the distance into consideration when calculating similarity, and finally fit a standard formula to calculate the similarity.(2) Short-time traffic flow forecasting. In this paper, we put forward an efficient method of short-term traffic flow forecasting:traffic flow forecasting based on ant colony clustering algorithm and RBF neutral network. As RBF neural network has the advantage in approximating capability, classification capability and learning speed etc., it is widely applied in all kinds of fields. As one of distributed optimization algorithms, ant colony algorithm shows its excellent capability of searching optimal solution, so it can be used in a variety of optimization problems. In this paper, ant colony algorithm is used for determining the centers. Besides, local search is used to improve the algorithm. We combine the improved ant colony clustering algorithm and RBF neural network to forecast the short-term traffic flow.