Please use this identifier to cite or link to this item: http://localhost:8081/xmlui/handle/123456789/6661
Title: VIBRATION BASED DAMAGE DETECTION IN STRUCTURAL SYSTEMS USING GENETIC ALGORITHM
Authors: Panigrahi, Soraj Kumar
Keywords: MECHANICAL INDUSTRIAL ENGINEERING;VIBRATION BASED DAMAGE DETECTION;STRUCTURAL SYSTEMS;GENETIC ALGORITHM
Issue Date: 2009
Abstract: The detection and identification of structural damage is a vital part of monitoring and servicing of structural systems. Engineering structures experience severe loads or impacts due to vibration of machines, earthquakes, winds, blasts etc. Most structures carry live loads, which may lead to fatigue failure. Undetected and un-repaired damage may lead to structural failure requiring costly repair or worse may lead to loss of human lives. So, damage assessment of theses structures is increasingly important in order to determine their safety and reliability. During the last few decades, vibration based methods have been developed to detect damage in civil, mechanical and aerospace engineering structures. These methods are based on the fact that the vibration characteristics of structures (namely frequencies, mode shapes and modal damping) are functions of the structural physical parameters such as mass, stiffness and damping. Structural damage usually causes a decrease in structural stiffness, which produces changes in the vibration characteristics of the structure. Sparsity and noise of measured data are inevitable in monitoring of engineering structures. Sparsity of measurements creates problem in identification of damage in structures / structural members. Most structures require a large number of degrees of freedom in their finite element models due to their size and complexity. But in practice, it is difficult and expensive to measure dynamic response at many locations. So, only a small subset of all the degrees of freedom in the model is normally measured. A number of schemes are available to deal with sparseness in modal data. But these methods are not always successful. Noise is also a big problem in identifying structural systems. The iii development of measuring skill and instruments may lower the noise level, but noise cannot be eliminated completely. An important fact about noise is its randomness. The experiment may be assumed to be deterministic if repeated identical experiments provide the same results. However, when all conditions under the control of the experimenter are maintained the same and the records still differ continually from each other, the process is said to be random. In such cases, a single record is not as meaningful as a statistical description of the totality of possible records. Therefore, without considering random noise in measurements, an identification algorithm cannot provide useful information about the structure. Many damage detection and assessment algorithms have been proposed to deal with noise polluted experimental data. Some have used artificial intelligence techniques for damage identification. Concept of residual force vector method along with Genetic Algorithm for damage identification in truss and beam structures have been studied by few authors. They have used complete modal information to develop the objective function. In case of beams, researchers tried to remove the rotational degrees of freedom by Improved Reduced System. From the literature review it appears that the minimum number of modal measurements required and their best location for identification of damage with reasonable accuracy has not been established. It is also important to know the number of natural frequencies and corresponding number of modal measurements required when experimental data is noise polluted. Therefore, the motivation for developing a new damage assessment algorithm arises naturally so as to provide a reasonable evaluation of damage with noisy and sparse measurements. Hence, the present work entitled 'Vibration based Damage Detection in Structural Systems Using Genetic Algorithm' has been attempted.
URI: http://hdl.handle.net/123456789/6661
Other Identifiers: Ph.D
Research Supervisor/ Guide: Mishra, B. K.
Chakraverty, S.
metadata.dc.type: Doctoral Thesis
Appears in Collections:DOCTORAL THESES (MIED)

Files in This Item:
File Description SizeFormat 
TH MIED G20521.pdf7.78 MBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.