|
Prestressed concrete continuous rigid frame bridge concrete bridge type is a structural integrity, reasonable force, concise and lively, while maintaining a continuous beam expansion joints, traffic smooth, there are T-rigid frame bridge with no bearing, convenient construction. At the same time, in the vertical loads, the main beam in the beam, the pier junction will produce negative moment, can make the main span positive moment value is smaller. In the long-term design practice, due to the structural analysis of the complex and lengthy, continuous rigid frame bridge structure optimization design mainly rely on the accumulated experience of the people, to the evolution of the way. This design process of a heavy workload, and empirical majority, so in the long span PC continuous rigid frame bridge design optimization study of the main parameters is necessary. National trunk road Hurong Hubei western part of the Longtan River Bridge is based on the project around the prestressed concrete continuous rigid frame bridge parameter optimization study, mainly to complete the following tasks: (1) to introduce domestic and foreign prestressed continuous rigid structure development status of the bridge, the construction of prestressed concrete continuous rigid frame bridge design parameters regression analysis to discuss the value of the experience of the various parameters. (2) the method of orthogonal experimental design to optimize the parameters of continuous rigid frame bridge. The results show that the orthogonal table parameter optimization to get better optimization results, get a combination of parameters: edge cross-ratio of 0.6, and the the beam bottom curve index 1.8, the main span and the the BEAMS high ratio of 19, thin-walled Pier Body The center spacing of 11.4m. And can ensure optimal accuracy of the premise, save optimization calculation time. (3) discussion of genetic algorithms, genetic algorithms and neural network combined with genetic algorithms, neural networks, and the orthogonal table combined with the efficiency of the three optimization methods. The results show that the simple genetic algorithm optimization, optimization very long time, and the results are as premature, it is difficult to get better optimization results. Full sample to train the neural network, good optimization results: side cross-ratio of 0.55, and the the beam bottom curve index 1.77 the cross high ratio of 18.96, double thin-wall pier center spacing of 11.37m. But calculating sample used a very long time, samples a week time. The complete orthogonal table to train the neural network, both to get good optimization results, they can calculate the sample time of day is a higher efficiency of the optimization calculation method.
|