E-bike Release Behavior Modeling And Organization Optimization

As a green way of travel,the electric bikes(e-bikes)have gained its popularity for its low price,low energy consumption,small size and travel convience,and become one of the main modes of our daily travel.However,there are also many associated disadvantages of the e-bikes in China,such as large flow rate,frequent traffic accidents,and management difficulties,which has drawn the attentions of various traffic researchers and practitioners.Due to the characteristics of small size,high speed and high density,the e-bikes' behavior is significantly different from that of motor vehicles or conventional bikes,which makes it difficult to directly apply the classical traffic behavior analysis methods to the e-bikes.In view of the above problems,the thesis focused in the e-bike flow,and utilized computer vision,physical modeling,mathematical simulation,etc.to investigate the release behaviors such as starting and expanding,and the corresponding organization optimization strategies of e-bikes.First,to solve the problems of large time consumption and low accuracy in the manual data extraction process of traffic video analysis,an automatic e-bike running data extraction method based on computer vision is proposed.The background initialization method is proposed with the stable sub-sequence clustering method,which can obtain the scene information of the traffic videos at the same time.Then,the datadriven background model is built to detect the e-bikes automatically.Finally,the detected e-bikes are tracked by the dense spatio-temporal context learning method,and the macroscopic traffic parameter representations of e-bike flow are established.Second,to investigate the microscopic interactions of e-bikes,a video based microscopic traffic behavior detection method is proposed by introducing the concept of temporal heat map generation.The changes of relative position and speed are simulated in the way of thermal accumulation and diffusion,which can recognize the speeding or traffic conflict behaviors of the e-bikes.Moreover,the e-bikes' traffic conflicts are further investigated,and the partial differential model is built to imitate the interactions and responses of the conflictive e-bike cyclists.The proposed numerical model can simulate the trajectories of conflictive e-bikes as well as imitating their spatial distributions.Third,the thesis focuses on the self-organizing characteristics of the release phase of the e-bike flow,and builds physical models to the e-bikes' starting behaviors and expanding characteristics.For the starting behavior of the e-bike flow,a series of quantitative studies are carried out based on the granular flow theories,such as Janssen's Law,Beverloo Equation,and particle dropping oscillation mechanism.The models can effectively simulate the starting characteristics of e-bike flow,which open up a new perspective for the e-bike traffic flow research.Then,the system entropy theory is introduced to model the expansion behavior of the e-bikes while entering the intersections.By simulating the time series of the position,speed and density entropy,the joint entropy distribution of the e-bikes is established,which can give a quantitative description of the self-organizing characteristics of the expansion of the e-bike flow.Finally,by summarize the previous conclusions on e-bike behavior mechanisms and characteristic analysis,the e-bike organization optimization methods are investigated.From temporal perspective,to reduce the interactions between e-bikes and adjacent motor vehicles,the impact factor of e-bike flow is established.With the e-bike early green strategy,the optimal timing method for the mixed traffic flow is carried out.From spatial perspective,the non-motor lane division problem is discussed in detail.Then,focusing on the trajectory offset characteristics of e-bikes,the “guide + constrain”channelization design is used to allocate the spatial resources in the intersections reasonably.By implementing this concept,the specific channelization design scheme is proposed for the left turn and straight direction of the e-bikes.In conclusion,the thesis has done a pioneer work to explore the mechanisms and characteristics of e-bike flow.The study will provide theoretical basis and method reference for the safety and organization perspectives of the e-bikes,which has decent social and economic significance and academic values.