Modeling Human-Structure Interaction Using A Controller System

The effects of human loads on structures are difficult to predict because they depend on the type of activity people are performing. However, models for typical activities such as standing, sitting and jumping have been proposed in the literature. Traditional models represent the human body as a system of lumped masses, dampers and springs arranged in a system with multiple degrees of freedom. Arguably, these models might not fully represent the human body because lumped masses, dampers and springs cannot add energy to the overall system. Furthermore, people could react differently to different levels of excitation and other environmental conditions. Controller systems have been widely used in electrical, seismic and other fields of engineering for systems in which setting a specific response is important. Given that the human acts like a controller system, where the feedback affects the response of the system, and the specific use of controllers is becoming common in structural engineering, this research developed a controller model to reproduce the phenomenon of Human-Structure Interaction (HSI). The methodology consisted in updating the parameters of the controller using experimental data from tests involving humans over a previously characterized structure. The controller system was the widely known Proportional, Integrative and Derivative (PID) controller and its derivations, PD and PI. Parameters of the controller were updated using a Bayesian probabilistic approach. Models were developed based on transfer functions obtained from experimental tests. A comparison of controller models and traditional Mass-Spring-Damper (MSD) models is performed at the end for validation purposes