A Study of Classification and Weight Determination of Moving Truck from Dynamic Bridge Strain by Using Scaled Model

Abstract

This research aimed to develop the classification and weight determination of moving truck by using only bridge strain response without axle detector or video camera to compute vehicle type from recorded photos. This was to reduce the cost and complexity of data analysis due to large number of measurement responses. The proposed system was capable to classify vehicle configuration from amount of axles and axle spacings. Bridge strain signals in each bridge section were modified to improve signal distinction and accuracy in the computation. Concept of correlation coefficient of strain responses between two sections of the bridge was employed to analyze moving speed, amount of axles and axle spacings. Based on the inverse problem of moving load and bridge dynamic response relationship, the system identified axle weight and gross weight using regularization least square with singular value decomposition (SVD) method. Moreover, to improve the accuracy of the analyzed truck weight, the updated static component (USC) technique was utilized in the final process. The study considered various truck type from 2-axle until 7-axle configurations type. Regarding to the experimental result from scaled model, it was found that the proposed system was able to successfully classify the truck configuration for all cases. The determination of truck gross weight allowed the average identification error of 3.75%. The maximum identification error was 8.58% at 95% confidence level.