In real traffic scenarios,real-time data acquisition and real-time processing are extremely critical issues,and the cur-rent traffic flow prediction has poor real-time performance,which is difficult to meet the needs of online analysis tasks.Based on this,real-time traffic prediction method combined with the Flink stream computing framework and the big data platform was pro-posed,which was based on the stream computing framework to capture and preprocess data in real time,including the use of Flink's transform operator to verify the data.The processed data sink to the BIG DATA HDFS file system,which was handed over to the next big data parallel framework for analyzing,modeling and training.The entire process of real-time traffic flow data inflow,pre-processing,and analysis modeling was simulated.Experimental results show that Flink can capture and preprocess traffic flow data in real time,and send the data into the distributed file system on time.On this basis,with the help of parallel analysis and modeling advantages under the framework of big data,it has a good effect on the real-time performance of data anal-ysis and prediction,which is better than the offline processing mode of GPU.
关键词
大数据/数据并行/流计算框架/实时处理/交通流预测/分布式系统/实时性分析
Key words
big data/data parallelism/stream computing framework/real-time processing/traffic flow forecasting/distributed systems/real-time analytics
维普
万方数据