用神经网络分析估计斜拉桥的施工控制参数

结合一座三塔四跨预应力混凝土斜拉桥施工过程,进行用神经网络分析估计斜拉桥施工控制参数的研究。用神经网络分析估计斜拉桥施工控制参数的过程主要包括建立神经网络、计算训练样本、训练神经网络和网络仿真四部分。运用神经网络仿真计算进行斜拉桥施工监控的具体方法是,先计算当前施工状态的索力和标高,再预测下一节段的索力和标高。经过往复循环,逐一进行节段预测调整,从而指导斜拉桥顺利施工。通过对比该斜拉桥合拢后部分节段索力及主梁控制点标高的计算值和实测值可知,用神经网络分析进行斜拉桥施工监控的效果达到设计的理想桥梁线型和索力要求。

Analyzing and Estimating the Parameters of Cable-stayed Bridge for Construction Control by Using Neural Network

Combining the construction process of a prestressed concrete cable-stayed bridge with 3 towers and 4 spans, research was carried out on analyzing and estimating the parameters of cable-stayed bridge for construction control by using neural network. The process for analyzing and estimating the parameters mainly covered 4 parts: the establishment of neural network, the calculation of training swatch, the training of neural network and the simulation of network. The specific method for cable-stayed bridge construction control by neural network simulating calculation is to first calculate the cable force and the elevation under current construction state and then estimate the cable force and the elevation of the next section. Through such repetitive cycles, each section is estimated and adjusted one by one to successfully guide the construction of cable-stayed bridge. After the final closure of the cable-stayed bridge, the calculated values of the cable force in some sections and the elevation for the control point of main girder are compared with on-the-spot measured values. It can be concluded that the effect for the construction control of cable-stayed bridge by using neural network analysis meets the requirement of the ideal alignment and cable force for the design bridge.