基于RS-BP融合的裂缝监控模型  

Crack monitoring model based on fusion structure of RS and BP

作　　者：雷鹏[1,2] 肖峰[2] 

机构地区：[1]长沙理工大学水利工程学院,湖南长沙410004 [2]中国水电顾问集团中南勘测设计研究院,湖南长沙410014

出　　处：《长沙理工大学学报（自然科学版）》2011年第2期45-50,共6页Journal of Changsha University of Science and Technology:Natural Science

基　　金：国家自然科学基金资助项目(50809025);湖南省教育厅科研资助项目(09C087)

摘　　要：目前对大坝裂缝的监控主要是应用统计模型,然而在实际应用中,其分析精度易受因子相关性的影响.本研究提出了基于RS-BP耦合的裂缝监控模型,该方法首先应用RS对裂缝监控信息进行属性和样本集约简,以提高网络的泛化分析能力;然后用BP神经网络对约简的样本集进行模式训练;最后根据训练好的网络对裂缝进行分析,以掌握裂缝的演变性态.计算结果表明,该方法计算效率和精度均较高,适用于分析裂缝等高度非线性问题.Crack is one of the main damages of concrete dams. Statistical model is the main monitoring model for crack nowadays, but its precision is easy to be interfered by associated factors in actual application. A new crack monitoring model was put forward based on weak coupling of rough set （RS） and back propagation neural network （BP-NN）. Firstly, RS was used to reduce samples and attributions of crack observation data so as to improve the analysis ability of NN. Secondly, the reduced samples were trained by BP-NN. Lastly, the trained NN was utilized to analyze the crack evolution and grasp its behavior. A case study showed that the proposed method is better in efficiency and precision, and is suitable for the analysis of highly nonlinear problems such as crack, etc.

关 键 词：粗集 神经网络 裂缝 监控模型 

分 类 号：TV698.1[水利工程—水利水电工程]