Bridge Structural Parameter Identification Based On Genetic Optimization Of Neural Network Algorithm

Bridge has already been the importance part of traffic foundation engineering in our country. However, due to the influence of many factors, the reinforced concrete bridge structures have always suffered some structure disease and damage, which are important incipient fault in traffic transporttation. Studying disease examination and deface evaluating technology of bridge structure is of far reaching importance in finding bridge structure potential defect timely, scientifically evaluating to impaired structure remnant longevity, guarantee bridge structure secureity and so on. Therefore, these research works become the current international hotspot at bridge engineering domain.Currently, the methods of bridge structure ability examination and structure evaluating are mainly based on apparent disease examination and static or dynamic load test. In this paper, based on the previous study results, combining neural network and genetic algorithm the segment flexural rigidity identification of reinforced concrete beam is studied by using the result of static load test. The main research findings are as follows:1. Based on the method of bridge structure ability examination and structure damage evaluation, the basic theory of genetic optimization neural networks algorithms is particularly expatiated. Genetic optimization neural networks algorithm is discussed in bridge structure parameter identification.2. The mechanics model of whole process of reinforced concrete beam section flexural rigidity is founded, based on the theoretical analysis the corresponding numerical calculation program is developed.3. In this paper, the mixed optimization algorithm which combine neural networks and genetic algorithms is put forward. The results show that the genetic algorithm has capability in fast learning of network weight, and the mixed optimization algorithm has capabilities in extensive mapping of neural networks and rapid global convergence of genetic algorithm. On this condition the realization program of genetic optimization neural networks is worked out, and the numerical examples are analyzed. Whether from the aspect of intuitive rational requirement or uncertain demand, optimization neural networks can effectively explain structure state changes, namely structure damage's location and degree.