Adaptive Localized Control of Structure-Actuator Coupled System Using Multi-Layer Neural Networks

Abstract

(Nakanarusawa-cho 4-12-1, Hitachi, Ibaraki, 316-8511, Japan)The identification and control of large-scale structures are considered to be difficult due to structural complicacy and system uncertainties. In this paper, based on the concept of localized control and neural networks, a computational algorithm for earthquake response control of large-scale structures is proposed. In this algorithm, the control-structure interaction (CSI) and actuator dynamics are considered. Moreover, electro-hydraulic actuators as practical applications are designed to simulate the proposed control algorithm. Results from the numerical simulations have shown great promise for the control of large-scale civil engineering structures with localized control using neural networks.

Direct identification of structural parameters from dynamic responses with neural networks

“…Among various neural networks with different topology structures, multi-layer neural networks are most commonly used in structural identification and control. Their applications to engineering problems have been summarized and reported in the literature (Ghaboussi and Joghatatie, 1995;Chen et al, 1995;Xu et al, 1999Xu et al, , 2000aWu et al, 2002;Nakamura et al, 1998;Yun and Bahng, 2000;Zhao et al, 1998).…”

A real-time structural parametric identification system based on fiber optic sensing and neural network algorithms

“…Numerous engineering applications of neural networks have been reported in the literature of recent years. A great number applications of neural networks for identification and control in civil engineering were reviewed by Ghaboussi et al [17], Chen et al [18], Xu et al [19][20][21][22], Wu et al [23], Nakamura et al [24], and Zhao et al [25]. Until now, most of the proposed methods for structural health monitoring damage with neural networks can be just used to give a qualitative indication or information that damage might be present in the structure, no quantitative identification can be determined [24,25].…”

Neural-network-based modeling of structural parametric evaluation with dynamic responses in time domain

“…Numerous engineering applications of neural networks have been reported in the literature of recent years. A great number applications of neural networks for identification and control in civil engineering were reviewed by Ghaboussi et al [5], Chen et al [6], Xu et al [7][8][9][10], Wu et al [11], Nakamura et al [12], and Zhao et al [14]. Until now, most of the proposed methods for structural health monitoring damage with neural networks can be just used to give a qualitative indication or information that damage might be present in the structure, no quantitative identification can be determined [12,13].…”