基于人工神经网络和遗传算法的大跨连续梁桥参数相关分析

主要论述了如何采用改进的BP神经网络和遗传算法进行现场施工参数的识别。为解决混凝土的容重、弹性模量所引起的误差,在丹江口二桥的施工控制过程中,采用3层BP神经网络进行混凝土的容重、弹性模量的识别;为确定预应力损失引起的标高偏差,引入了遗传算法对其建模分析,取40个初始染色体群,以5个世代繁衍不再出现更优的染色体作为终止GA计算的条件。文章以这两种方法在丹江口二桥施工过程的预测控制分析中的成功运用为实例,证实了神经网络控制理论和遗传算法在连续梁桥的施工过程的预测与控制中的实用性和有效性。

Analysis of Parameters for Long-span Continuous Bridges Based on Neural Network and Genetic Algorithms

The exposition centered on how to apply the improved BP Artificial Neural Network and Genetic Algorithms in the identification of the field construction parameters.The improved triple-layer BP Artificial Neural Network is applied to identificate concrete density and elasticity modulus to solve the density of concrete and the error caused by elastic modulus in the Second Bridge of Danjiangkou construction control process.The Genetic Algorithms is introduced in the modeling analysis of the prestressed losses to find the error of the elevation.It takes that 40 initial chromosome groups that never appear better chromosomes after five generational multiply as the condition of stopping GA calculation.The functionality and validity of these two methods applied in the preview and control of the construction of continuous girder bridge are verified through a project named the Second Bridge of Danjiangkou,located in Hubei province of China.