摘要
基于人工神经网络并结合土石坝渗压监控物理模型,采用遗传学习算法对神经网络BP模型的初始权重进行优化,建立了土石坝渗压预测仿真BP模型,对土石坝的渗透压力及时间等特殊因素进行计算机仿真。在此基础上将这些知识信息分布于各神经元的连结点,实时显示土石坝渗透压力变化状况,为土石坝安全监测管理提供了一种理想的资料分析模式。实例应用山西省文峪河水库的历史资料进行训练、预测和计算机仿真,结果表明,该算法收敛速度较快,仿真精度较高,为土石坝安全监测系统的资料分析提供了一种新思路和新方法。在大坝安全运行和管理中,意义重大。
Based on the artifical nerval network and combined to the physical model of the osmotic pressure monitoring for embankment dam, Genetic Algorithm was used to optimize initial Nerval weight of Network BP model, to establish the forecast simulation BP model of the osmotic pressure of embankment dam, and to proceed computer simulation for osmotic pressure of embankment dam, time and some especial factors. On the basis of this, these knowledge information was distributed to every neurons junction to real-time display movement status of the darn. That is an ideal data analysis pattern to supply safety monitoring management of the embankment dams .As an example, training, forecast and computer simulation were proceeded employing the historical data of the Wenyu River reservoir in Shanxi Province. As a result, this kind of arithmetic was provided with obvious rapidness constringency speed and clear highness simulation precision. This research may give the data analysis of the dam safety monitoring system to bring a new way. So, it is signality for the safety run and management of the dam.
出处
《系统仿真学报》
CAS
CSCD
北大核心
2008年第4期1052-1055,1059,共5页
Journal of System Simulation
基金
国家高技术研究发展计划(863计划)(2004AAA001050)
山西省自然科学基金(2006011059)相关子项目
关键词
土石坝
神经网络
计算机仿真
安全监控
embankment dams
nerval network
computer simulation
safety monitoring
