Pedestrian-Induced Load Identification from Structural Responses Using Genetic Algorithm with Numerical and Experimental Validation

Abstract

A reliable pedestrian load model is a prerequisite for the accurate calculation of human-induced structural vibration. In recent years, many researchers have proposed pedestrian load models based on direct force measurements, such as force plates. However, direct measurement techniques often suffer from incapability when applied to real structures in operation and can hardly be used to measure crowd loads. In this paper, an inverse load identification method is proposed to extract pedestrian vertical load from structural responses. Through a genetic algorithm, the pedestrian’s pacing frequency, dynamic load factors, and phase angles in the Fourier-series model are identified from structural acceleration responses. The proposed algorithm is further investigated for the identification of multiple-pedestrian load parameters, where structural displacement responses are used to give an equivalent number of pedestrians. Numerical examples demonstrate that the pedestrian load parameters are estimated with high accuracy and robustness against noise and modeling errors. A sensitivity analysis is given to explain the different estimation accuracies among the parameters. Finally, the proposed method is validated through an experimental test, showing its practicality for identifying pedestrian loads in real structures.