Construction Control Technique For Long-span Continuous Rigid Frame Bridge Based On Improved BP Neural Network

With the rapid transit development of China's highway and railway, more long span bridges were needed to span great rivers and bays, PC continues rigid frame bridge emerged as the times require and was developed fleetly in recent years. The continues rigid frame bridge has a property of bond of beam and abutment, so not only maintain the advantage of comfort of continue girder, but also decrease maintenance cost because that there is no bearing of "T" beam. Continue rigid frame bridge is more and more widely used because of its high strength, lightful line type, convenient and fast construction, long-span ability.The prestressed concrete continuous rigid frame bridge often adopts the economical and reasonable segmental construction technique with cantilever installation. During the process of the balanced cantilever construction, the bridge structure is influenced by various factors, such as shrinkage, creep and temperature change of concrete, construction error, survey error, and so on. Such factors result in the complex change of the internal forces and displacements in the bridge. The bridge construction control is essential measure and step to ensure the quality and safety in the procedure of the bridge construction. In this essay the technique of the structure calculation and error adjustment measures used in bridge construction prefigurative deviation control was researched based on the construction control and monitor project of Yunnan Bao-Long freeway Shatian river bridge.Through the collection of long-span prestressed concrete continuous rigid frame bridge construction control data, with the help of the existing research achievement, the paper pursues for quality of the common forecasting method, and raises the method of construction control, based on the theory of BP Neural Network. BP algorithm is actually a nonlinear optimal problem and produces inevitably local minima. The paper proposes the methods that combination algorithm of the additional momentum and variable learning-rate.At last via the data comparison with the practical survey data draw a conclusion: It is feasible for the construction prefigurative deviation control and error adjust used in PC rigid frame bridges to use the calculation model and program brought forward in this paper.