Artificial intelligence techniques for reliability assessment of nonlinear bridge systems

Artificial Intelligence techniques are used to develop simulation-based methods for the reliability evaluation of complex and large structural systems with application to highway bridge structures.; Load intensity, bridge response, and structural capacity are uncertain in nature. The object of structural reliability theory is to account for the uncertainties encountered during the load capacity evaluation of structural components and systems and to account for these uncertainties during the calibration of safety factors for design codes. In order to perform a reliability analysis, two tools are required: the first tool is an accurate nonlinear analysis program capable of analyzing the behavior of structural systems for a specific (deterministic) set of loading conditions and material properties. The second tool is a systematic search algorithm that will identify the probabilistically dominant failure modes accounting for the randomness in applied loading and material properties.; The purpose of this study is to develop the two tools necessary to evaluate the safety of bridge structural systems. The first tool consists of new analysis models for steel and prestressed concrete highway bridges that are implemented in a modified version of the program NONBAN. The new models accurately represent the nonlinear behavior of bridge members using empirically derived information from moment-rotation curves and plastic hinge length equations. The second tool consists of a new Shredding Genetic Algorithm (SGA) that is designed to efficiently identify dominant failure modes of bridge structural systems. SGA improves the traditional genetic algorithm by interfering with the natural selection process and filtering out the weakest and unhealthy offsprings using the principle of elitism.; To improve the efficiency of the analysis process, a new solver, referred to as the pseudoforce method, is derived based on the Sherman-Morrison-Woodburg formula. By combining SGA with the pseudoforce solver, the reliability analysis of bridge systems is reduced by a factor of 4 compared to classical search methods in combination with traditional analysis techniques. The proposed techniques will eventually lead to better control of the safety of bridge structural systems and improve the reliability of structural designs. Examples are provided to demonstrate the efficiency and applicability of the proposed models.