چكيده لاتين
Abstract
The existence of crack in structures is considered one of the types of damage that many methods, including theoretical and non-theoretical methods, have been introduced to detect them. Due to the complexity of modeling the crack line in the structure, those theoretical methods that require analysis of the cracked structure for detect of cracks, are limited. In this thesis, for the first time a model-based non-theoritical method that is a combination of Peridynamic theory and an upgraded metaheuristic optimization algorithm is proposed for crack detection in steel plates and beams with the aim of simple modeling, relatively high accuracy in modeling the cracked structure, highly accurate and reliable results, and also high convergence speed.
Unlike conventional methods, the Peridynamic theory has high capability of modeling the crack in structure and is used to analize the cracked structure. Also, the charged system search algorithm (CSS) is chosen in the crack search step. In order to increase the accuracy and reliability of the results of this algorithm, a special search step has been added to it (UCSS). Unlike many previous studies, there is no limitation in the direction of the crack in this research. The objective function in crack identification problem is defined based on the difference in displcament at different points of the structure. Before implementation of crack detection method, the optimal position of the displacement recording sensors as well as the external loads applied, are determined. At first, the mensioned method is evaluated to detect cracks in a rectangular plate in different scenarios. The method is then tested on new scenarios in which the crack orientation is similar to the previous scenarios but the crack length is reduced. Next, the method is applied to several beams with different restraint conditions. In order to more investigation of power of added step in the proposed method, four other optimization algoritms search the crack in beams. In the next section, crack detection in bending plates is also investigated. Finally, several examples of crack detection solved by one of the thesis references with another method, are considered. The noise effect on crack identification method in examples of beams and plates is considered .To reduce the computational cost of proposed crack detection method, the parallel processing capability is used in the implementation of the structure analysis part.
The results obtained by the crack detection method show that the coordinates of crack line founded by this method in plates and beams have a very good agreement with the real crack and the crack detection in noisy environmental conditions is also acceptable. The search power of the UCSS algorithm has also been higher compared to the mentioned algorithms. The proposed method is also very accurate in identifying cracks defined in membranes whose length is 10%. The identification accurate of crack coordinates using the proposed method in this thesis on examples provided by one of the references was also higher than the methods used in that reference.
Keywords: Crack detection, membranes and plates, Peridynamic theory, optimization algorithm
