Structural damage evaluation using genetic algorithm

Abstract

The detection and identification of structural damage is important in monitoring of structural systems during their lifetime. Many researchers have proposed a variety of damage evaluation methods based on structural monitoring. The stiffness matrix is used in some conventional damage detection methods; however, it leads to inevitable error due to the lack of data provided by structural monitoring. To overcome this problem, this study introduces a new damage evaluation method that identifies the structural damage in a shear building based on a genetic algorithm using the structural flexibility matrix with dynamic analyses. The proposed method enables the deduction of the extent and location of structural damage, even when there is insufficient data on the dynamic characteristics and insufficient accurate measurements of the structural stiffness and mass. The validity of the proposed damage evaluation method is demonstrated through numerical analyses using OpenSees.

Introduction

The detection of structural damage is essential in ensuring structural safety during a structure's lifetime. The objective of structural damage detection is to evaluate the quantitative and qualitative deterioration of the structural system in service or under a severe load. From both safety and performance viewpoints, it is desirable to monitor the occurrence, location, and extent of deterioration. Structural deterioration results from internal factors such as corrosion, fatigue, and aging, and from external factors such as earthquakes, wind loads, and impact loads. To localize structural damage, conventional damage detection methods, such as non-destructive testing (NDT) inspection, are commonly adopted in field tests; however, excessively vulnerable areas cannot be tested using NDT. The largest disadvantage of NDT is that it can only be used when the damaged area is clearly visible or vulnerable areas are easily locatable. Therefore, for a more effective approach to damage detection, a globalized inspection method, rather than a localized one, is essential.

Recently researchers have proposed a variety of damage detection methods based on structural response monitoring. Particular attention has been paid to structural damage detection methods using natural frequencies and mode shapes, calculated from the structural responses without accurate information regarding the material properties of the structural system [1], [2]. These early studies using natural frequencies and mode shapes, however, have a number of problems. As a result of the incompleteness of the natural frequencies and mode shapes derived through measurements, a larger amount of measurement data is required to accurately assess damage and some modes cannot be captured via testing due to the limited number of sensors. Accordingly the use of conventional forward analyses enables detecting of the existence of structural damage, but not the location or extent of that. To overcome this problem, mathematical approaches have been proposed in an effort to minimize the difference between the analytical responses and the measured responses. One approach is to formulate an optimization problem for the equation and output error [3]; another is to update the structural system using the system identification (SID) for more precise simulations of the current structural responses. The SID approach includes the optimal matrix modification method [4], the minimum rank perturbation method [5], and the Eigenstructure assignment technique [6]. Recently damage detection through a mode analysis that improves upon these earlier approaches has begun to be used widely [7], [8]. In particular, some research is being undertaken on the use of the flexibility matrix, which uses both natural frequencies and mode shapes [9], [10], [11].

Accordingly research is also being conducted on damage detection methods that can be used in structural health monitoring through structural responses. A variety of damage detection methods, including artificial neural networks [12] and genetic algorithms [13], are used to obtain global optimal solutions, because these methods are not sensitive to the initial value and originate from the laws of nature. The use of an artificial neural network requires a neural training phase, for which a large amount of data is required. In contrast, the genetic algorithm does not require a large amount of data: this algorithm is based in the laws of natural evolution and survival. An important characteristic is seeking a global optimal solution using multiple point routes rather than a single point route. Consequently damage detection based on genetic algorithms is significant in searching for structural damage.

From these viewpoints, this paper proposes an improved detection method for structural damage based on a genetic algorithm using a flexibility matrix for shear. The proposed method assesses damage by calculating a flexibility matrix that only uses the lower modes of structures instead of structural forward analyses based on changes in the natural frequencies and mode shapes determined through modal analyses. Accordingly it is possible to detect the extent and location of structural damage, even in the absence of information about the structural stiffness. The feasibility and applicability of the proposed method were verified through numerical analyses using an open source program (OpenSees).