Parameter Sensitivity Analysis And Elevation Control On Long-span Continuous Rigid Frame Bridge

Due to continuous rigid frame bridge construction is small adjustable process and affected by some certainty and uncertainty factors, meanwhile it has more construction stage in cantilever construction,coupled with structural system of the conversion during construction process, it is necessary to control the construction process of continuous rigid frame bridge. For continuous rigid frame bridge, elevation control is the top priority. In practical engineering, variation amplitude of stress is large, but elevation measurement can be controlled within the accuracy. From the previous practice experience, if elevation reaches target state, the structure of the internal force will tend to achieve the corresponding control limits.This paper studies the following:Firstly, this article summaries continuous rigid frame bridge, and development situation of construction control. It also expounds main influencing factors in the construction process, compares and analyses several prediction methods, and analyses the basic theory of BP neural network in detail.Secondly, through establishment of finite element model of continuous rigid frame bridge with practical project, it can conduct parameter sensitivity analysis in construction control. Sensitivity parameters range between about -10% and+10%, then select control objectives, such as maximum deflection under the condition of cantilever girder calculating the amplitude girder changes under control target. According to influence degree of the parameters on control target is different, we can determine the sensitivie parameters and the non-sensitive parameters.Thirdly, it discusses vertical mold elevation of continuous rigid frame bridge detailed, and determines the elevation control scheme of MeiShan bridge.Then analyses deflection monitoring results and fold segment of pushing fold on T-shape structure in detail.Fourth, Using BP neural network, elevation prediction model can be established Compared with other prediction methods, BP neural network model can achieve good results on elevation control of continuous rigid frame bridge.Taking left pieces of Meishan Bridge for example, closed height is only 2 millimetres when both ends close. Adjacent segment elevation error of the difference beams does not exceed 15 millimetres, so it reached the bridge bottom line smooth purpose.The largest difference is 26 millimetres from the target elevation.Overall, a lignment control of the bridge construction has achieved good results.