Road Traffic Congestion Forecast Based On Big Data

The construction of a powerful transportation country is conducive to residents’ travel and promotes social and economic prosperity and development.Future urban traffic road congestion prediction based on artificial intelligence algorithms can help traffic decision-making departments to plan traffic flow in advance and avoid traffic congestion;real-time urban road congestion statistics based on big data technology can dynamically reflect the law of current traffic congestion,It is convenient to take measures such as adding tidal lanes to remove congestion,and at the same time,it can respond to traffic congestion caused by temporary road repairs,traffic accidents and other potential emergencies in a timely manner.These have important practical significance for road traffic congestion prediction.This paper mainly studies future urban road congestion prediction and real-time urban road statistics,predicts future road travel time based on gradient boosting algorithm,and uses big data real-time computing engine to perform urban road prediction statistics based on real-time data stream to solve road traffic congestion.The main work of this paper is as follows:1)Aiming at the autoregressive problem of future travel time predictions,a variety of algorithm fusion missing value interpolation processing methods and the XGBoost time series prediction algorithm based on feature influence features are proposed.The algorithm dynamically splits the residuals to find the maximum Gain improves algorithm accuracy,integrates pre-training simulation,linear regression,median fitting,etc.to optimize the missing value processing process,performs weighted feature extraction,uses Bayesian optimization,cross-validation for parameter optimization design,and avoids over-fitting Together,the complexity of the model is reduced,and the stability of the model is improved.Experimental results show that the algorithm can quickly converge,achieve the purpose of short training time and strong generalization ability,and has a good prediction effect for the autoregressive problem of future travel time prediction.2)Aiming at the problem of real-time urban mass vehicle distribution and deduplication,a dynamic expansion multi-bit allocation mass data deduplication algorithm is proposed.The algorithm receives vehicle information in each road area of the entire city,and distributes massive vehicle data evenly into an N-dimensional matrix through multiple allocations to confirm the road area where the vehicle is located in real time.After accumulation,the traffic density of each road area is obtained,and the bitmap is used Realize dynamic expansion.Experimental results show that the algorithm can solve the problem of deduplication of mass vehicle data distribution under the premise of allowing extremely low errors.3)In view of the analysis of the causes of road congestion problems,a multi-dimensional real-time urban traffic congestion prediction statistical model based on Flink was designed and implemented.The model abstracts three dimensions.According to the real-time vehicle data information on each road in the city,it is consumed by Kafka,using non-deterministic finite state automata to customize complex events,pattern matching of speeding vehicles,outputting dangerous driving vehicle information and warning,avoiding causes Traffic congestion caused by traffic accidents;using window function calculations to obtain the congestion situation of urban roads;and in order to count the most congested roads in all road sections in the entire city in real time,a custom TopN function is introduced.At the same time,it solves the problem of time disorder of traffic incidents caused by problems such as network delay.The experimental results show that the use of water level lines can solve the problem of incident time disorder to the greatest extent and feed back the real road traffic conditions.Similarly,Flink can complete the traffic caused by traffic accidents caused by urban vehicle speeding while taking into account real-time and throughput.Congestion and the most congested road statistics.