Study On FCD Based Traffic Parameters Detection And Applications In Traffic Condition Analysis

Accurate parameters detection and road traffic condition analysis are the foundation of modern traffic management, control, guidance and planning. At present, in the field of traffic parameters detection, the crucial problems of GPS–FCD(GPS-Floating Car) based detection technique, as a new direction of traffic detection, is how to improve the performance of detection. On the other hand, it is insufficient to reflect the road traffic condition depending on road traffic parameters only, how to construct analysis and estimation methods that correspond the practical situation, has been paid wide attention.In this paper, the project is mainly conducted base on the key detection technique of traffic parameters from GPS–FCD and road traffic condition analysis. For floating car traffic parameters detection, the key points are breaking through the foundation technique problems existed in road layer correction and data pre-processing field, to research the method that improves the performance of road average velocity estimation using mixture floating cars (bus, taxi etc). In road traffic condition analysis, discussing the inherent law of vehicles flow state through traffic parameters acquired by floating car, from the macroscopic viewpoint; researching the change and its law using microcosmic following characteristic of vehicles driving from the microcosmic viewpoint.The main research in this paper as follows:①Making full use of the abundant information implied in floating cars data, road layer correction model for road nodes through large scale data analysis to solve the key problem in business map accuracy correction has been established. Firstly, the paper adopts a method of road nodes matching based on road network topological structure to revising the non-normal nodes in original road layer. Secondly, constructs the static buffers on the basis of roadway nodes, meanwhile, using GPS data to construct dynamic buffers of nodes based on static buffer. Finally, the paper chooses the GPS points in the dynamic buffers to guarantee accuracy of clustering data maximally. a method for road nodes accurate information determine is proposed based on hierarchical cluster using these data, and correction of road nodes is implemented. The effectiveness of this method has been tested through actual road experiment.②For different type of floating car (bus, taxi etc), a method is constructed in this paper, it aims at specific zero data pre-processing respectively. On the basis of this method, a method to improve estimation performance of road segment average velocity is put forward by means of integrating floating car data from single-car to multi-car.For the case that GPS data usually have masses of zero velocity data, on the basis of bus and taxi driving properties and characteristics of GPS data, zero data processing algorithms is proposed for bus and taxi respectively. Furthermore, a preprocessing model taking floating car zero velocity data into account is established to solve the delay processing problem caused by floating car unusually stop at station, crossing or in run-time.According to different situations of relative position among floating car data series and start-stop nodes and preprocessing model proposed above, to construct the single car road segment average velocity model for single class car based on local driving time respectively, and forms the multi-car road average velocity model for single-class car taking the proportion that the valid data of this car dominate in all valid data of same class cars as weighting. On the other hand, multi-car road segment instantaneous average velocity model for single-class car has been established by means of hierarchical cluster integrate the instantaneous time series of single car on road segment; two kinds of road segment average estimation model above-mentioned has been integrated, and constructs road segment average velocity model; Finally, according to the different features of bus and taxi, blend floating car road segment average velocity estimation model based multi-cars established take consideration of proportion of bus valid data dominate in taxi valid data. Model has been tested through practical data from floating car.③The paper proposes a method for traffic discrete-time series analysis based on the Detrended Fluctuation Analysis (DFA), and provides a new technique to reveal inherent law of car flow status with variation traffic parameters. In most circumstances traffic flow parameters obtained from practical detection are unstable and limited; the paper analyzes traffic flow time series from the viewpoint of scale invariance, and demarcates the index of road segment average velocity of taxies and buses by introducing the Detrended Fluctuation Analysis (DFA). The paper discusses the long-range correlation and internal regularity of the time series of traffic flow using scaling exponents of actual traffic flow time series. The result indicates that the road segment average velocity of buses exits long-rang correlation, but taxies doesn't.④This paper proposes the method for road traffic condition change and its laws from microcosmic car following characteristics, and discusses the stability of road traffic state by mean of building a new car following model, and then analyzes the road traffic state change and its laws.The change of road traffic condition is an intrinsic objective reflection of vehicle driving behavior. The development of modern sensor, communication and Internet of Vehicles technique provide advantages for depicting microcosmic features of vehicle driving. A extended Tow Car-following (TCF) model and a Backward Looking & Multiple Car-following (BLMCF) have been put forward respectively for description of the following features of vehicles driving on road accurately and grasp the change regularity of road traffic condition. This paper also discusses the distribution of vehicle spacing and velocity varying with time in road traffic flow, based on the models provided above. The simulation results show that it not only contributes to improving the stability of road traffic condition, but also restrains the occurrence of congestion phenomenon in road traffic, and improves the road traffic flow status while considering the effects on following vehicle from adjacent vehicles and leading vehicles which take backward looking effect into consideration in road traffic system effectively.From the above, the contributions in this paper are: proposes a road layer correction method facing with road nodes based on floating car techniques, an estimation method for road segment average velocity based on blend floating car, an analysis method for traffic time series based on the DFA, and an analysis method for traffic state change regularity using car following characteristics. The effectiveness of the work has been tested and verified by experiment results.