Study On Urban Road Traffic State Identification And Spatio-Temporal Characteristics Based On Different Types Of Floating Car Data Fusion

Real-time identification and spatio-temporal characteristics analysis of urban road traffic state is the key to grasp the operation of urban road traffic, which can provide support and reference for traffic guidance, management and planning. Accurate analysis of the state of urban road traffic is one of the effective measures to ease the current traffic congestion problems. A huge number of GPS data accumulated by floating vehicles has many advantages, such as wide area covering, high real time, low maintaining cost and high reliability compared with dynamic traffic data from fixed traffic detectors. These GPS data can reflect changes in the process of road traffic states. However, there is a lack of effective research on using different types of GPS floating car data for urban traffic state analysis. As for spatio-temporal state characteristics analysis for traffic flow in road network, the current spatio-temporal autocorrelation analysis system can’t better reveal inherent laws of evolution in different periods or between adjacent sections.Therefore, this paper is mainly determined from the following two aspects: real-time road traffic state identification using two different types of floating GPS data and spatio-temporal state characteristics analysis for traffic flow in road network, with a view to improving the current state identification method of urban road traffic and revealing traffic network inherent laws. Research work of this paper includes the following three aspects:First, this paper confirms the urban traffic state identification standards based on dynamic clustering. Considering that current traffic state division exists fuzziness and actual state of the road traffic is just subjective information, this paper uses massive floating car GPS data and the improved K- means clustering algorithm to determine the best types of traffic state divided number. Then the fuzzy C-means clustering method is introduced to evaluate the state of road traffic and finally the threshold of different traffic state is obtained, which can be the criterion of real-time traffic state identification in Chapter 5.Secondly, this paper proposes an urban road identification method based on different kinds of floating car data fusion. Combining the advantages and disadvantages of two kinds of floating cars and taking into account of buses and taxies operational characteristics and road network coverage, road average speed estimation based on bus GPS data and taxi GPS data is respectively put forward. Then the average speed is fused from the feature-level in order to make it more accurately reflect the continuous changes of traffic state. Using real data to test the proposed identification method, experimental results show that this method has good effect of identification with high real-time feature.Thirdly, this paper analyses spatio-temporal state characteristics for traffic flow in road network. The improved K- means clustering algorithm proposed in Chapter IV is used for traffic state clustering of each spatio-temporal unit, and the space-time network connectivity is defined to achieve quantitative analysis of traffic state accumulation. Traffic relative congestion degree and relative volatility are further used to analyze spatio-temporal state characteristics for traffic flow in road network. The experimental results have shown the effectiveness of the analysis.In summary, this paper presents an urban road traffic identification method based on different kinds of floating car data fusion and analyses spatio-temporal state characteristics for traffic flow. Experimental results have verified the effectiveness of the above research, which can provide support and reference for urban road transport services and management.