Estimation Of Expressway Abnormal Events And Traffic State Information Based On Characteristic Parameters

The basic traffic parameters are the basis of traffic state estimation research,while the characteristic traffic parameters are the key indicators to characterize the traffic state in a specific traffic scenario.In the past,most studies on traffic state used basic traffic parameters,which lacked the relevance for specific scenarios such as episodic congestion,and it was difficult to comprehensively and effectively characterize the traffic state such as congestion degree and traffic capacity under the scenario.Selecting or designing the corresponding characteristic traffic parameters for specific scenarios enriches the means of traffic state characterization and makes up for the shortage of traffic state estimation studies in different scenarios.Therefore,the traffic state estimation method established by using the characteristic traffic parameters under abnormal events is of great value to improve the highway service quality and alleviate traffic congestion.The paper takes the selection and design of characteristic traffic parameters as the entry point,and designs the congestion type discriminative algorithm for the difference between episodic events and frequent events.Based on the influence of each traffic factor on the capacity of the incident point under episodic congestion,a capacity estimation model at the bottleneck is established,and the theoretical foundation and data support are laid for the establishment of the incident time and incident location estimation model by combining the diffusion behavior of the traffic event influence.The main work is as follows.(1)Design congestion judgment and frequent and episodic congestion discrimination algorithms based on hierarchical clustering.In order to solve the problem of "dimensional disaster" caused by the poor applicability of existing congestion discrimination algorithms to different road sections and the high dimensionality of clustering,this paper designs suitable feature parameters based on the mutation theory and spatial correlation state for the differences of different congestion types.Based on the top-down hierarchical clustering process and the FCM algorithm,we design the congestion judgment and the algorithm to discriminate between frequent and episodic congestions.The effectiveness of the algorithm is verified by comparison test with Mc Master algorithm.(2)Neural network-based capacity estimation algorithm.To address the problems of poor applicability of capacity estimation algorithms in existing literature,inability to capture the change of capacity at the early stage of an event,and failure to consider the intrinsic correlation among traffic factors,this paper selects characteristic traffic parameters to characterize the capacity by analyzing the degree of influence of traffic factors on the capacity.By building a BP neural network model,the mapping relationship between each traffic factor and the capacity is established.The experimental results show that the algorithm of this paper has better effect compared with the discount factor algorithm,KNN algorithm and SVM algorithm.(3)Incident time and incident location estimation model.The congestion time and location detected by the existing detection algorithms have some spatial and temporal errors with the incident time and location.Accordingly,this paper establishes the estimation model of incident time and incident location based on(1)(2)points,considering the diffusion behavior of the impact of abnormal events,and the experimental results show that this algorithm has a high estimation accuracy.