چكيده به لاتين
The main purpose of this thesis is to provide an automated image-based system for the detection and analysis of cracks in concrete structures, including bridges. The main core consists of a combination of two automatic systems, "crack detection" and "crack type classification". A combination of fuzzy classifiers has been used to design these system as an intelligent agents in the framework of "multi-agent system". These agents are related to each other and exchange information. The optimization of these systems has been done by the particle swarm optimization method. The results of validation of test data set show that the accuracy of classification by the proposed systems is satisfactory and they shows high reliability in crack image analysis. Another important task is the automatic quantification of different types of cracks in concrete. For this purpose, in this dissertation, an image-based method is presented that in addition to the usual characteristics of cracks, can accurately and automatically measure other characteristics such as the exact width of the crack in all points, location and maximum amount of crack width, crack direction, number and area of meshes and the number of cracks belonging to a crack group. The main idea of this part is the simulation of cracks based on graph theory and matching the index points and crack segments with graph nodes and edges. In general, the condition of any type of defect can be judged by the main characteristics of the severity, extent, and type of defect. On the other hand, the detection and analysis of cracks as one of the most effective types of defects is of particular importance in assessing the condition of concrete structures. In this thesis, a new (crack-based) condition index is presented to assess the structural health of reinforced concrete bridges. This index is designed as a function of the three main factors of severity, extent, cause (type) of cracking and also the importance degree of structural members. By optimizing the grading values of these coefficients, this index can be used to estimate the overall condition of concrete bridges as a more accurate criterion with greater reliability.The most important innovations and achievements in this thesis include providing an intelligent automatic method for detection and classification of cracks in concrete members, a method of automatic analysis and quantification of all crack types with high accuracy and speed, and presenting a new structural health condition index for concrete bridges.
