Study On Pre-processing Methods Of Traffic Data Based On Space-time Relative

Currently data era has slowly into our work and life, and this is big data analysis platform for urban development policy-making brings visual expression. Big Data platform is an all-encompassing data warehouse, in order to make the data contained in the process of being analyzed to obtain as more accurate results as possible all the data needs to go through a washing and screening process; thus transport large data platform as the most important type of data set, which is also subject to this inevitable process. Data monitoring system to monitor the road to get traffic plays a vital role in project design and evaluation, the accuracy of its data monitoring statistics are directly affect the future direction of policy-making. The current road traffic monitoring system mainly uses a detection coil and video surveillance equipment to analyze road traffic data. Due to an aging equipment and software failure, the statistical results often occur error, which led to the data for the study based on other analyzes cannot be performed or error analysis. Therefore, a pretreatment process traffic data has great significance.This paper analyzes the common data anomalies recognition methods, and analyzes their advantages and disadvantages from the perspective of the applicability and operational feasibility. As a basis for a period of time and traffic data with a spatial characteristic features and lane departure, it proposed based on the time associated with traffic data correlated with spatial correlation combining space-time preprocessing methods. In the time-dependent data traffic, in order to make the data in the identification and repair process to fully rely on the data itself cycle characteristic, while the data identifying the end with a variable scale adjustment, and therefore the introduction of time-series data having a multi-scale variable analysis of wavelet transform and wavelet transform theory to fourth-order Newton- Cotes theorem rational use greatly reduces the computational difficulties. In the spatial correlation of data traffic respect, from the point of view of the road adjacent lane utilization, with its different adjacent lane traffic flow distribution as the basis for the identification of data anomalies, while the adjacent lane spatial relationship data repair process. Finally, this paper use Liangqing Nanning fifth avenue and Golden Elephant Avenue East intersection imported three straight lane as study example, the amount of data traffic in the middle lane straight conducted to identify anomalies and repair, and to evaluate the effect of the error analysis as a repair indicators, for unprocessed data case, only to repair the results of wavelet transform case, wavelet transform and combine adjacent lane repair fix the spatial relationship case. From the analysis results, the combination of the results of the time-dependent wavelet transform and spatial correlation of adjacent lane repair is significantly superior to the results of a single treatment of repair.