Short-term traffic flow prediction method based on space-time feature selection and Kalman filtering

Abstract

The invention discloses a short-term traffic flow prediction method based on space-time feature selection and Kalman filtering. According to the method, time sequences with different time delays of adjacent intersections are used as input characteristics of the model, and the defect that a traditional model cannot fully consider space-time correlation between traffic flow time sequences is avoided. The method comprises the following steps: firstly, carrying out flow aggregation on original SCATS data; then, applying a multidimensional scaling method to the prediction sections and the section groups in the adjacent areas of the prediction sections, and finding out some sections with high correlation with the prediction sections; performing feature selection on space-time features formed bydifferent time delays of the sections, and determining the optimal input feature; and finally, obtaining a prediction result through a Kalman filtering model considering space-time correlation. The model can obtain high prediction precision when short-time flow prediction is carried out on the urban intersection, and has good robustness.

Description

technical field [0001] The invention relates to the technical fields of Kalman filter algorithm and traffic flow prediction, and in particular, a short-term traffic flow prediction method based on spatio-temporal feature selection and Kalman filter. Background technique [0002] The Intelligent Transportation System (ITS) is based on the research of key basic theoretical models, with the purpose of alleviating road congestion and reducing traffic accidents, organically combining advanced information technology, electronic control technology, data communication technology and electronic positioning control technology. Then establish a comprehensive, wide-ranging, high-efficiency, intelligent transportation analysis and route selection management system. The analysis and prediction of road short-term traffic flow is the key content of intelligent transportation system. Carry out targeted research on short-term traffic flow prediction theory and methods, and obtain effective an...

AI Technical Summary

Problems solved by technology

In the first type of model, the linear regression model has the disadvantage of poor real-time performance. The historical average model is relatively simple, but the prediction accuracy is not high, and it cannot handle unexpected accidents; although the second type of model has high prediction accuracy, it takes a long time to predict. The shortcomings of cumbersome parameter adjustment are obvious; the third type of model updates the model through real-time traffic flow to ensure the real-time prediction, but requires a large amount of historical data for training, which is difficult to meet in practice; the fourth type of model satisfies the Requirements for forecasting accuracy and real-time performance, but combined forecasting models are difficult to apply in practice

In addition, the existing models still have the defect of not fully considering the temporal and spatial correlation of traffic flow

Images