Using search based optimization algorithms in Bridge Weigh-In-Motion systems

Abstract

A method for vehicle axle load identification on slab-on-girder bridges is presented in this paper. The method's basis lies in developing a bridge specific static influence line matrix and using an optimization method combined with a pattern search algorithm. A 1/3 scale laboratory model of a six girder bridge was used as part of the case studies demonstrating methodology development and implementation. A finite element bridge model was developed and calibrated against experimental data. The search based optimization procedure was needed to estimate vehicle characteristics when noise was present in the recorded data and when error was present in the numerical model. The actual vehicle and the estimated vehicle characteristics therefore did not exactly match - a deviation designated as an identification error. The identification error's statistical behavior with respect to the level of noise in the measured response or error in the model was studied. Methods for identification error reduction were numerically evaluated. The importance of a field calibrated bridge model was shown, as was that averaging multiple vehicle estimates for different positions on the bridge can decrease the identification errors.