Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 82
PROCEEDINGS OF THE EIGHTH INTERNATIONAL CONFERENCE ON THE APPLICATION OF ARTIFICIAL INTELLIGENCE TO CIVIL, STRUCTURAL AND ENVIRONMENTAL ENGINEERING
Edited by: B.H.V. Topping
Paper 27

Optimal Fuzzy Control of Hybrid Base Isolation System using Genetic Algorithms

H.S. Kim+, P.N. Roschke* and D.G. Lee+

+Department of Architectural Engineering, Sungkyunkwan University, Suwon, Korea
*Department of Civil Engineering, Texas A&M University, College Station, TX, United States of America

Full Bibliographic Reference for this paper
H.S. Kim, P.N. Roschke, D.G. Lee, "Optimal Fuzzy Control of Hybrid Base Isolation System using Genetic Algorithms", in B.H.V. Topping, (Editor), "Proceedings of the Eighth International Conference on the Application of Artificial Intelligence to Civil, Structural and Environmental Engineering", Civil-Comp Press, Stirlingshire, UK, Paper 27, 2005. doi:10.4203/ccp.82.27
Keywords: hybrid base isolation, MR damper, neural fuzzy, fuzzy logic control system, genetic algorithms.

Summary
This study investigates performance of a genetic algorithm (GA) optimized fuzzy logic controller (FLC) for a hybrid base isolation system consisting of a friction pendulu system (FPS) isolator and a magnetorheological (MR) damper. It has been shown that hybrid base isolation can protect a structure from extreme earthquakes without sacrificing performance during more frequent, moderate seismic events. Because of the inherent robustness and ability to handle nonlinearites and uncertainties, FLC is used in this study to operate a large MR damper which is a key component of the hybrid base isolation system. Although FLC has been used to control a number of structural systems, selection of acceptable fuzzy membership functions has been subjective and time-consuming. To overcome this difficulty, a GA is employed to optimize FLC for the control of a hybrid base isolation system [1]. The GA applied in this study focuses on finding appropriate fuzzy control rules as well as adjusting the membership functions. To this end, the FLC is designed using a GA with a local improvement mechanism (Nagoya approach) [2]. The Nagoya approach utilizes mechanisms of genetic recombination in bacterial genetics. It is efficient in improving local portions of chromosomes. However, the number of fuzzy rules that makes a FLC should be decided by the designer of the control system before applying this method. Sometimes an appropriate number of fuzzy rules cannot be readily selected. Therefore, in what follows, a weighting factor associated with each rule is introduced into the chromosomes in order to let the GA weaken or strengthen the contribution of each rule. Root mean squared (RMS) structural accelerations and base drifts that are normalized with respect to the uncontrolled RMS acceleration and drift responses, respectively, are used as the objective functions as well as normalized peak acceleration and drift responses. Moreover, a weighted sum approach is introduced to combine multiple objectives into a single fitness function. The level of priority of control for structural accelerations and base drifts can be adjusted by varying the weighting factors used in the fitness function.

The proposed design approach using the GA-optimized FLC for a hybrid base isolation system is demonstrated with the help of numerical simulations. Parameters from a large scale experimental model are employed as the basis for the numerical simulation. The large scale experimental test was conducted at the National Center for Research on Earthquake Engineering (NCREE) in Taipei, Taiwan. Although the expert's knowledge-based FLC controls the hybrid base isolation system effectively in comparison with passive control strategies during the experimental test, there seems to be considerable room for improvement through the use of an optimal design method. Therefore, this experimental model of a hybrid base isolation system is employed as a numerical example in order to demonstrate improved performance of the FLC by using the proposed design approach. The powerful modeling capabilities of adaptive neuro-fuzzy inference systems are used to develop a neuro-fuzzy model of the MR damper and the four FPSs that support the mass. Modeling of the FPS is carried out with a nonlinear analytical equation and neuro-fuzzy training. Behavior of the FPS and MR damper can be successfully estimated using these neuro-fuzzy models.

A passive damping strategy, human-designed FLC and conventional semi-active controller (i.e. skyhook) are used to compare the efficiency of the proposed GA-optimized FLC based on computed responses to several historical earthquakes. In the passive-on control case, base drift can be significantly reduced but structural acceleration is not well controlled. The skyhook controller reduces structural acceleration in comparison with passive-on control, but only at the expense of larger base drifts for all earthquakes that are numerically simulated. A human-designed FLC can reduce base drift better than the skyhook approach and it can reduce structural acceleration better than passive-on operation of the MR damper. That is, a human-designed FLC can appropriately control both base drift and structural acceleration. Finally, a GA-optimized FLC shows better performance in comparison with the human-designed FLC. Furthermore, performance of the GA-optimized FLC can be easily adjusted by selecting an appropriate weighting factor according to desired performance requirements. Based on numerical studies, a hybrid base isolation system consisting of a large MR damper and a novel FPS with an appropriate controller is shown to achieve significant decreases in base drift without accompanying increases in acceleration that accompany passive base isolation systems.

References
1
C.L. Karr, L. Freeman, D. Meredith, "Improved fuzzy process control of spacecraft autonomous rendezvous using a genetic algorithm", SPIE Conf on Intelligent Control and Adaptive Systems, 274-283, 1989.
2
T. Furuhashi, Y. Miyata, K. Nakaoka, Y. Uchikawa, "A new approach to genetic based machine learning and an efficient finding of fuzzy rules", Proc of the 1994 IEEE World Wisemen/Women Workshop of Fuzzy Logic and Neural Networks/Genetic Algorithms, 114-122, 1994.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £80 +P&P)