Research on Collection and Preprocessing Strategies of Traffic Data Driven by Big Data

In this paper, the microwave data of traffic are collected by RTMS instrument, and the traffic data are preprocessed with the help of standardized timestamp and anomalous data processing techniques, and the spatio-temporal correlation of traffic data is analyzed. On the basis of the spatio-temporal characteristics of traffic flow, the traffic flow missing data repair model (TDIM) based on the third-order tensor Tucker decomposition is constructed to realize the repair of traffic flow missing data. The results show that there is an obvious change cycle law between the traffic data of urban road network and the road itself, and there is mutual influence and propagation phenomenon between the traffic conditions of different road sections. The combination of spatio-temporal correlation can be used to construct a more comprehensive and accurate model for repairing traffic data. In addition, in three different datasets, the data restoration errors obtained by this paper’s algorithm are reduced by more than 25% compared to the LRMC, PPCA, LLS and KSR-EN methods, and -5.16%-4.47% compared to the HaLRTC algorithm, which shows that this paper’s method has the lowest traffic data restoration errors.