Construction of traffic volume estimation models based on the multiple regression analysis of waveform features extracted from bridge dynamic monitoring data

Abstract

This study is aimed at constructing the traffic volume estimation models. To construct the prediction model, the bridge acceleration data and traffic volume of target bridge were acquired. The multiple regression analysis was adapted as the method of constructing the prediction models. Measures related to dynamic response waveform were used as the features of this study and to be considered for explanation variables. By a procedure based on the correlation analysis, the suitable variables were determined and the estimation models for ten-minute total traffic volume and the truck traffic volume were constructed. From those results, the constructed estimation model showed good agreement with the traffic volume data.