上海交通大学学报

上海交通大学学报（自然版） ›› 2014, Vol. 48 ›› Issue (11): 1655-1659.

• 其他 • 上一篇    下一篇

海堤渗压神经网络分布模型的建立及规律分析

黄铭1，刘俊2   

  1. (1. 合肥工业大学 土木与水利工程学院， 合肥 230009；2. 上海交通大学 船舶海洋与建筑工程学院， 上海 200240)
  • 收稿日期:2013-09-27
  • 基金资助:

    国家自然科学基金项目(50979056)资助

Establishment and Analysis of Artificial Neural Network Distributed Model for Sea Wall Osmosis Pressure

HUANG Ming1,LIU Jun2   

  1. (1. School of Civil Engineering, Hefei University of Technology, Hefei 230009, China；2. School of Naval Architecture, Ocean and Civil Engineering, Shanghai Jiaotong University, Shanghai 200240, China)
  • Received:2013-09-27

摘要：

摘要：  为有效揭示海堤渗压的分布特征，掌握其在临海工作环境下的特殊规律，在利用神经网络建模优点的同时，采用多测点渗压监控信息，并将测点坐标因素加入到输入层，综合前期潮位因子、积分型降雨因子、时效因子，形成海堤渗压神经网络安全监控分布模型结构，以实测信息进行建模训练计算；在获得合理训练结果基础上，根据输入层因子补充插入坐标样本，获得不同位置的渗压模型值及渗压分布曲线.文中以广义回归神经网络为例，结合浦东海堤实测资料，以实例说明以上述方法在神经网络不提供显式的情况下，建立可获得分布曲线的监控模型，并以此对海堤渗压分布规律特色加以分析.

关键词: 海堤渗压, 神经网络, 分布模型, 影响因子, 坐标

Abstract:

Abstract： In order to describe the distributed character of sea wall osmosis pressure  and analyze its rule under special coastal condition, artificial neural network was considered with multiple survey point information. Coordinates of survey points were added to input layer together with former tidewater factor, integral rain factor and time effect factor. The monitored data were used to train the distributed model. Based on the training result, complementary coordinate samples were put into the model, and osmosis pressure of different locations and distributed curve were obtained. The general regression neural network was used, taking the monitored information of Pudong sea wall as instances. The instances illustrated the establishment of distributed model and the obtainment of distributed curve, although the neural network does not give apparent expressions. Moreover, sea wall osmosis pressure special distributed rules were analyzed using these results.

Key words:  sea wall osmosis pressure, artificial neural network, distributed monitoring model, effect factor, coordinate

中图分类号: