BP神经网络在连续梁桥施工监控中的应用

Application of BP Neural Network in the Construction Control of Continuous Beam Bridge

为了实现大跨度连续梁桥施工过程中立模标高快速、准确地确定,基于BP神经网络能逼近任意的函数与自适应算法结合的特点,将其运用到连续梁桥施工控制的标高预测中.通过有限元软件建立桥梁模型,结合参数的影响的分析,运用BP神经网络原理,根据实测值与理论值的对比分析结果来确定挠度预测的输入向量和目标向量,建立大桥高程偏差的神经网络模型.利用MATLAB程序的神经网络模型,完成对样本矢量的输入及对桥梁施工控制的网络训练,预测出下一阶段的标高值,以此反复进行,有利于立模标高更快更精确的确定,最终使桥梁的线形和设计线形达到很好的吻合.

In order to fast and accurately determine the construction process of large span continuous beam bridge formwork elevation, according to the characteristics that BP neural network can approximate any function and adaptive algorithm based on the combination, we made use of prediction model to the construction control of continuous beam bridge. We established the bridge model by the finite element software, analyzed the impact of binding parameters, then by using the principle of BP neural network, according to the results of analysis to determine the input vector and the target vector deflection prediction are compared with theoretical values measured, we set up the neural network model of elevation deviation. By using the neural network model of the MATLAB program, the sample vector and the input of the bridge construction control network training were carried out, and the next phase of the elevation value was predicted, then repeating again and again, which is helpful for the formwork to be elevate faster and more accurately, and the bridge shape and design line shape to achieve good agreement.