Large-Scale Computational and Experimental Analysis and Design of Smart Control Systems
Extreme events such as earthquakes and hurricanes have caused significant damage to structural and infrastructure systems. Hazards impacts on structural integrity and functionality of these systems continue to pose significant threats to public safety and regional and national economies around the world. An objective of RAMSIS lab is to develop smart structural systems through utilization of active and semi-active controllers and sensing instruments.
Development and Large Scale Experimental Validation of New Optimal SMC-Based Control Algorithms:
In two research projects supported by Korea Research Institute for Smart Materials and Structures, we developed new control strategies for mitigation of seismic-induced poundings in multi-span bridges using semi-active Magnetorheological (MR) dampers. Specifically, a new semi-active control algorithm called SMC-OPC was developed that is based on a clipped sliding mode control (SMC) with sliding surfaces designed using an optimal polynomial control (OPC) approach. We also developed a novel control strategy involving fractional order sliding surfaces called fractional sliding mode control (FSMC) method in order to integrate desirable features of sliding mode control with fractional order calculus concepts. The control designs use a stochastically linearized model of the nonlinear bridge with passive components of the magnetorheological dampers embedded in the system model to achieve a more representative system characterization. We conducted a series of large-scale shake table tests at Pusan National University in South Korea on a 6.5 ton three-span bridge supported by rubber and lead rubber bearings. In these experiments, our proposed optimal control algorithms yielded noticeable improvements over other control methods.
Enhanced Stochastic Averaging of Non-Integrable Nonlinear Systems Subjected to Stochastic Excitations:
Various control methods in their design process rely on models describing initial system behaviors or use stochastic linearization to simplify nonlinear models of systems. However, the accuracy of these methods in representing system behaviors when operating under stochastic excitations reduces as the nonlinearity in the system increases. We have addressed this limitation by developing an “enhanced stochastic averaging method of energy envelope” that generates an equivalent nonlinear stochastic model for non-integrable systems. This has been achieved by treating such systems as diffusive Markovian processes with a transition probability density function governed by the Fokker-Planck-Kolmogorov (FPK) equation. The drift and diffusion components of the FPK equation are derived by applying stochastic calculus. We have also derived an equivalent excitation intensity and modified damping parameters by equating drift and diffusion components of the modified and the original system through the method of weighted residuals using high order moments of system velocities. Validation studies for a seismically excited building on a nonlinear foundation against Monte Carlo simulations revealed the high accuracy of the proposed method in analytical estimation of the probability distribution function of the energy of the system.
Reliability‐Based Control Algorithms for Nonlinear Hysteretic Systems:
New reliability-based control methods have been developed by utilizing enhanced stochastic averaging o energy envelop to analytically formulate limit state functions of nonlinear systems in the objective function of control algorithms. Proposed methods have the potential to considerably improve the reliability of controlled systems, as has been demonstrated via extensive numerical investigations, and offer an optimal cost-effective design and upgrade solution for various structural and infrastructure systems.