CN-122024488-A - Method, system and storage medium for predicting traffic flow

CN122024488ACN 122024488 ACN122024488 ACN 122024488ACN-122024488-A

Abstract

The application relates to the technical field of intelligent traffic, and discloses a method, a system and a storage medium for predicting traffic flow. The method comprises the steps of obtaining multisource dynamic data of a mountain road network, constructing a road network connection relation model, extracting flow transfer characteristics and transfer dynamic fluctuation characteristics among nodes, carrying out time sequence trend analysis on a historical data sequence of a flow transfer path, predicting a transfer influence range caused by road network state change, screening an affected road subset when the transfer influence range exceeds a preset range threshold value, generating a flow transfer spatial distribution map, updating weight parameters of the road network connection relation model according to the transfer influence range, carrying out flow offset simulation, determining the adjusted road network connection state characteristics, carrying out geographic space mapping on the road network connection state characteristics, extracting traffic flow quantification indexes, and outputting traffic flow prediction results when preset congestion risk conditions are met. The method and the system can improve the accuracy and pertinence of the prediction of the traffic flow of the mountain road network.

Inventors

PU DANDAN
LIU RANRAN
ZHANG WENCHUAN
CHENG HAIKUN
CHENG ZHIBIN
WANG PENGFEI
WANG TIANYI
WANG XIAOYU
GUO YU
WANG XIN

Assignees

郑州市交通规划勘察设计研究院有限公司

Dates

Publication Date: 20260512
Application Date: 20260325

Claims (10)

1. A method for predicting traffic flow, the method comprising: s1, acquiring multisource dynamic data of a mountain road network, constructing a road network connection relation model, extracting flow transfer characteristics among nodes from the road network connection relation model, and extracting transfer dynamic fluctuation characteristics based on the change condition of the flow transfer characteristics in a time dimension; s2, carrying out time sequence trend analysis on the historical data sequence of the flow transfer path, and predicting a transfer influence range caused by road network state change by combining the transfer dynamic fluctuation characteristics; S3, screening an affected road subset based on a road network topological structure when the transfer influence range exceeds a preset range threshold, and generating a flow transfer spatial distribution diagram according to a transfer flow density index corresponding to the road subset; s4, updating weight parameters of the road network connection relation model according to the flow transfer spatial distribution diagram, performing flow offset simulation, and determining the adjusted road network connection state characteristics; s5, performing geospatial mapping on the road network connection state characteristics, extracting traffic flow quantification indexes based on a spatial mapping result, and generating a visual distribution map of traffic flow evolution; And S6, judging whether the traffic flow quantification index meets a preset congestion risk condition, and if so, outputting a traffic flow prediction result.
2. The method of claim 1, wherein in S1, the step of obtaining multi-source dynamic data of the mountain area road network and constructing a road network connection relation model includes: acquiring longitude and latitude coordinates and traffic state data of a mountain road network through a satellite image and a ground sensor network, and defining the longitude and latitude coordinates and traffic state data as multi-source dynamic data; Noise filtering and time stamping are carried out on the multi-source dynamic data, and preprocessing data are obtained; Identifying flow transfer paths according to a time-dependent change sequence of longitude and latitude coordinates in the preprocessing data, determining transfer probability distribution according to historical occurrence frequencies of each flow transfer path, and determining initial time delay distribution corresponding to each flow transfer path according to a time difference value in the change sequence; Generating a mountain area road network dynamic data set based on the preprocessing data, the transition probability distribution and the initial time delay distribution; And according to the mountain road network dynamic data set, adopting a graph structure analysis method to model the connection relation between road nodes and road connection edges of the mountain road network, and obtaining the road network connection relation model.
3. The method according to claim 2, wherein in S1, extracting a traffic transfer feature between nodes, extracting a transfer dynamic fluctuation feature based on a change condition of the traffic transfer feature in a time dimension, comprises: Based on the road network connection relation model and transition probability distribution corresponding to each flow transition path, determining a state transition relation among nodes, and constructing a flow transition probability matrix as the flow transition characteristic; Setting a sliding time window, extracting time sequence data of transition probability among nodes in the flow transition probability matrix in the sliding time window, calculating the statistical standard deviation of the time sequence data, and obtaining the transition dynamic fluctuation characteristics.
4. The method of claim 1, wherein S2 comprises: Acquiring a historical data sequence corresponding to the flow transfer path; Carrying out trend analysis on the historical data sequence by adopting a time sequence trend test method, and determining the trend direction and trend significance of flow transfer change in the historical data sequence by combining the transfer dynamic fluctuation characteristics; Based on the trend direction and the trend significance, potentially affected road nodes or road segments associated with the traffic diversion path are determined, and a diversion impact range caused by road network state change is predicted accordingly.
5. The method of claim 1, wherein S3 comprises: Judging whether the transfer influence range exceeds a preset range threshold, if so, performing topology search in the transfer influence range based on a road network topology structure represented by the road network connection relation model, and screening out an affected road subset; extracting transition probabilities among nodes in the road subset according to the flow transition characteristics, and constructing a local flow transition probability matrix corresponding to the road subset; determining the transfer flow of each road section in the road subset based on the local flow transfer probability matrix and the current traffic flow of the road subset, and calculating the transfer flow density index corresponding to each road section by combining the traffic capacity coefficient of each road section; And performing space mapping on the topological space positions of the road subsets and the transition flow density indexes to generate the flow transition space distribution map.
6. The method of claim 1, wherein S4 comprises: Determining the passing resistance increment of each road section in the road subset and the blocking frequency of each node based on the flow transfer spatial distribution diagram; Updating weight parameters of connecting edges of corresponding nodes in the road network connection relation model by combining the passing resistance increment and the blocking frequency; performing flow deviation simulation based on the updated road network connection relation model to obtain flow distribution deviation of the road network under the flow transfer condition; and determining the adjusted road network connection state characteristics according to the flow distribution deviation.
7. The method of claim 1, wherein S5 comprises: Combining the geographic position coordinates of the corresponding road sections and preset area influence weights, and performing geographic space mapping on the road network connection state characteristics; extracting traffic flow quantification indexes according to the space mapping result, wherein the traffic flow quantification indexes at least comprise average traffic duration and path delay increment; and generating a visual distribution map reflecting the dynamic evolution process of the road network flow based on the spatial mapping result.
8. The method of claim 1, wherein S6 comprises: based on the traffic flow quantification index, calculating the comprehensive congestion risk index and the traffic flow saturation of the corresponding road section; Judging whether the comprehensive congestion risk index and/or the traffic flow saturation exceeds a corresponding preset congestion risk threshold, if so, predicting the predicted traffic flow and congestion spreading trend of the corresponding road section in a future preset period based on the spatial mapping result and the road network connection state characteristic; And taking the predicted traffic flow and the congestion propagation trend as the traffic flow prediction result.
9. A system for predicting traffic flow for implementing the method of any one of claims 1 to 8, the system comprising: The feature extraction module is used for acquiring multisource dynamic data of a mountain road network, constructing a road network connection relation model, extracting flow transfer features among nodes from the road network connection relation model, and extracting transfer dynamic fluctuation features based on the change condition of the flow transfer features in a time dimension; The prediction analysis module is used for carrying out time sequence trend analysis on the historical data sequence of the flow transfer path and predicting the transfer influence range caused by road network state change by combining the transfer dynamic fluctuation characteristics; The screening mapping module is used for screening the affected road subset based on the road network topology structure when the transfer influence range exceeds a preset range threshold value, and generating a flow transfer spatial distribution diagram according to the transfer flow density index corresponding to the road subset; The iteration adjustment module is used for updating the weight parameters of the road network connection relation model according to the flow transfer spatial distribution diagram, carrying out flow offset simulation and determining the adjusted road network connection state characteristics; The space mapping module is used for carrying out geographic space mapping on the road network connection state characteristics, extracting traffic flow quantification indexes based on space mapping results and generating a visual distribution map of traffic flow evolution; and the prediction output module is used for judging whether the traffic flow quantification index meets a preset congestion risk condition or not, and if so, outputting a traffic flow prediction result.
10. A computer readable storage medium having instructions stored thereon, which when executed by a processor implement a method for predicting traffic flow according to any one of claims 1 to 8.

Description

Method, system and storage medium for predicting traffic flow Technical Field The application relates to the technical field of intelligent traffic, in particular to a method, a system and a storage medium for predicting traffic flow. Background Along with the development of intelligent traffic systems, traffic flow prediction technology is widely applied to scenes such as road operation monitoring, traffic organization optimization, trip induction, congestion early warning and the like. The early stage of traffic flow prediction research mainly relies on an empirical model based on historical statistical rules, and the estimation of the traffic state in the future period is realized by carrying out statistical analysis on parameters such as the road section flow, the vehicle speed, the occupancy and the like. Then, along with the development of sensor networks, positioning technologies and multi-source data acquisition technologies, traffic flow prediction gradually evolves from single section data analysis to multi-source traffic information fusion, and traffic states can be modeled by comprehensively utilizing road network topological structures, vehicle track data, geospatial data and real-time monitoring data. In recent years, with the development of graph models, time sequence analysis methods and space association analysis methods, traffic flow prediction further develops towards networking, dynamic and refinement, traffic changes of a single road section are focused, and the transfer and propagation rules of traffic flows between adjacent nodes and roads and the influence of local traffic disturbance on the running state of the whole road network are focused. The existing traffic flow prediction method has a certain application effect in road environments with strong regularity such as urban plain road network, but has stronger dynamic property and uncertainty in the transition path and evolution process of traffic flow in the road network due to complex road trend, more gradients and curves, obvious fluctuation of traffic conditions, easiness in influence of factors such as weather, geological conditions, local congestion propagation and the like for mountain road network scenes. The existing method is usually focused on trend extrapolation of a target road section based on historical flow data or conventional state prediction based on a static road network structure, so that traffic flow transfer characteristics among nodes in a mountain road network, transfer dynamic fluctuation characteristics and an effect relationship of the traffic flow transfer characteristics on expansion of an influence range are difficult to effectively describe, and meanwhile, the integrated modeling, mapping and prediction mechanisms are also lacked for spatial diffusion of traffic flows in a subset of affected roads, road network connection state change and congestion risk evolution processes, so that a prediction result is difficult to accurately reflect the dynamic evolution process of the mountain road network traffic state. Therefore, how to construct a traffic flow prediction method capable of accurately predicting a traffic flow dynamic change process according to mountain road network scenes and integrating multisource dynamic data, road network connection relations, traffic flow transfer characteristics, spatial distribution evolution and congestion risk states becomes a key technical problem to be solved in the field. Disclosure of Invention The application provides a method, a system and a storage medium for predicting traffic flow, and aims to solve the technical problems that in the prior art, the traffic flow dynamic change process under a mountain road network scene is not characterized sufficiently, the transfer and propagation rule of traffic flow is difficult to reflect accurately, and the traffic flow prediction cannot be performed by effectively combining road network connection state change, so that the prediction accuracy and the applicability are low. In a first aspect, the present application provides a method for predicting traffic flow, the method comprising: s1, acquiring multisource dynamic data of a mountain road network, constructing a road network connection relation model, extracting flow transfer characteristics among nodes from the road network connection relation model, and extracting transfer dynamic fluctuation characteristics based on the change condition of the flow transfer characteristics in a time dimension; s2, carrying out time sequence trend analysis on the historical data sequence of the flow transfer path, and predicting the transfer influence range caused by road network state change by combining with the transfer dynamic fluctuation characteristic; S3, screening an affected road subset based on a road network topological structure when the transfer influence range exceeds a preset range threshold, and generating a flow transfer spatial distribution diagram according to a