Research On Urban Road Traffic Status Discrimination And Prediction Method Based On Multisource Data Fusion

The identification and prediction of urban road congestion is an important component of urban modernization construction.How to fully utilize the existing diverse urban traffic data,mine traffic operation patterns,and make more accurate judgments and predictions of road conditions,so that the public can obtain more accurate traffic information and make reasonable travel plans,has strong practical value in alleviating urban road traffic congestion.From the current research status,the fusion of multi-source road traffic information data can make reasonable and full use of the data.This article conducts research on the discrimination and prediction methods of urban road traffic status based on multi-source data information.The main work is as follows:(1)Acquisition and processing of multi-source data.The paper first introduces the basic structure of floating vehicle data,road network structure,points of interest,and internet road conditions,and conducts preliminary processing to improve the accuracy and credibility of the data.Then,the preprocessed floating vehicle trajectory data is matched with the road network data to obtain the trajectory of the vehicle on the road section during driving.Finally,based on the relevant characteristics of traffic flow parameters,the average speed of the road segment based on the floating vehicle data is extracted from the floating vehicle data,which,together with other traffic flow related parameters,serves as an important reference for evaluating road traffic status.(2)Perform multi-source data fusion.Due to measurement and calculation errors,the speed data provided by different data sources all have a certain degree of noise interference.To address this issue,this paper proposes a speed data fusion method based on long-term and short-term memory networks to fuse the speed data.Through experimental verification,the fusion speed combines the characteristics of multiple different sources of speed and comprehensively utilizes the traffic information contained therein,resulting in a higher spatiotemporal coverage compared to single source speed and a better comprehensive reflection of traffic operation status.(3)Establish a prediction model for urban road traffic flow based on multi-source data.A neural network model for predicting urban road traffic flow parameters based on multi-source data fusion is proposed.The focus is on the optimization of the LSTM neural network using the improved Sparrow Search algorithm based on Logistic chaotic mapping.Traffic flow parameter prediction models are established on the basis of the LCSSA-LSTM neural network model,in order to obtain the prediction results of traffic flow parameters and provide a data basis for traffic state prediction.In addition,comparative experiments were conducted on traffic flow prediction before and after integrating weather factors,indicating that integrating weather factors can improve the accuracy of traffic flow parameter prediction to a certain extent.After experimental verification,this model can achieve good results in predicting traffic flow parameters.(4)Construct a road traffic state discrimination and prediction model based on GA-FCM algorithm.On the basis of fuzzy mean clustering algorithm,aiming at the problem that it is greatly affected by the selection of initial value and prone to local convergence,this paper proposes to use genetic algorithm to optimize the selection of initial value of fuzzy mean clustering,and constructs a traffic state discriminative model based on GA-FCM.Use this model to cluster and analyze traffic flow parameter data,in order to achieve the division of urban road traffic operation status.Then,combined with the predicted traffic flow parameter data,the traffic status can be predicted.