Small Sample Traffic Flow Forecast Method Based on Beetle Antennae Search Algorithm and Support Vector Regression

Short-term traffic status refers to the traffic status information with a time interval of no more than 15 minutes. The accurate short-term traffic state prediction information can help traffic managers better control and coordinate vehicles, and also provide key traffic information for drivers' intelligent driving. However, the commonly used prediction algorithms often need large data samples to support, but in some sections which could not provide a big sample of traffic data or lack big data, the prediction accuracy of these algorithms will be greatly reduced. Based on the small amount of data of short-term traffic flow in some sections, combined with the fast search speed of Beetle Antennae Search algorithm and the high accuracy of Support Vector Regression algorithm in the case of small samples, this paper proposed a fast and accurate small sample traffic flow prediction model. Finally, the measured traffic flow data collected by the PEMS system in California were selected. After reasonable pretreatment of this data, this paper used the BAS-SVR model to output forecast results, the final test results showed that BAS-SVR had an excellent prediction effect in a small sample of data.