Urban perception data processing method based on space-time causal relationship

Abstract

The invention provides an urban perception data processing method based on a space-time causal relationship. The urban perception data processing method is used for repairing vehicle space-time trajectory behavior data. The method comprises steps of learning the spatial correlation of the trajectory acquisition equipment through a skip graph model, and outputting equipment distributed vector representation for subsequent processing; taking the noise data detection problem as a sequence marking problem, detecting error data and potential missing data through a bidirectional LSTM sequence marking model, marking vehicle track points, and predicting missing track points by using a bidirectional LSTM-based sequence predictor; and finally, correcting wrong track points in combination with the predicted missing data. According to the method, the noisy data is automatically detected, and the missing data is repaired from the perspective of the space-time causal relationship, so that the unreasonable vehicle trajectory is prevented from being generated, and the correct rate of wrong trajectory data repair is improved.

Description

technical field [0001] The invention belongs to the technical field of vehicle spatiotemporal trajectory behavior data processing, and in particular relates to a noise processing method for urban perception data based on spatiotemporal causality. Background technique [0002] Urban perception is the basis of urban computing. By deploying different types of sensors in different geographic spatial locations, continuous and collaborative monitoring of the natural and human environments of urban areas can be achieved. With the rise of sensing technology, various types of spatiotemporal data are collected by geospatial sensors, such as traffic flow data collected by inductive loop detectors and remote traffic microwave sensors, communication data collected by base stations. In addition, the application of multi-sensors to generate spatio-temporal data in reality also includes meteorological monitoring, electrical equipment monitoring, weather forecasting, environmental status mon...

AI Technical Summary

Problems solved by technology

[0009] In view of the problems of uneven distribution, low sampling rate and high failure rate of sensors in the current intelligent transportation system, the large-scale vehicle trajectory behavior data collected is noisy and poor in reliability, and manual error correction is time-consuming and labor-intensive. The present invention provides A method for urban sensing data processing based on spatio-temporal causality is proposed to automatically detect noisy data and repair missing data from the perspective of spatio-temporal causality to avoid generating unreasonable vehicle trajectories

Images