改进补偿模糊神经网络在坡面产沙模拟预报中的应用
CFNN Modification and Its Application in Predicting Soil Erosion on Slope Land
作者：代华龙(四川大学 水力学与山区河流开发国家重点实验室，四川 成都 610065)；曹叔尤(四川大学 水力学与山区河流开发国家重点实验室，四川 成都 610065)；耿新宇(西南石油大学 计算机科学学院,四川 成都 610500)
Author：(State Key Lab. of Hydraulics and Mountain River Eng., Sichuan Univ., Chengdu 610065, China)；(State Key Lab. of Hydraulics and Mountain River Eng., Sichuan Univ., Chengdu 610065, China)；(School of Computer Sci.,Southwest Petroleum Univ., Chengdu 610500, China)
收稿日期：2008-02-22 年卷（期）页码：2008,40(5):45-50
期刊名称：工程科学与技术
Journal Name：Advanced Engineering Sciences
关键字：补偿模糊神经网络；侵蚀产沙；坡面
Key words：compensatory fuzzy neural network; soil erosion;slope
基金项目：国家自然科学基金资助项目(50739002)
中文摘要
要：土壤侵蚀产沙是一个极其复杂的物理过程，其物理机理十分复杂，往往难以用数学方式来描述。根据土壤侵蚀过程内在特点，采用一种改进的补偿模糊神经网络(CFNN)方法，对坡面侵蚀产沙过程进行了预测研究。利用四川某水土保持实验站观测数据，通过MATLAB语言编制程序进行了仿真试验, 全局误差为0.0517，预测值与实测值吻合良好，误差小于10%。结果表明，该方法预报精度较好,为坡面产沙模拟模型的建立提供了重要的参考方法。
英文摘要
Soil erosion is a very complicated physical process and its physical mechanism is very complex so that it is difficult to describe by mathematical expression. Compensatory fuzzy neural network (CFNN) is a complex system of compensatory fuzzy logic (CFL) and artificial neural network (ANN). It has higher systematic stability and error tolerance. Base on the dynamic and physical behavior of soil erosion, the present paper modified the CFNN to predict soil erosion on slope land. MATLAB computer program had emulated using the observed data at somewhere Station of Water and Soil Conservation in Sichuan province. The system error was 0.0517, and difference between predicted and the observed values was less than 10%. The results indicate that the modified CFNN is much more precision than traditional artificial neural network. It provides a vital tool for soil erosion modeling and predicting.
【关闭】
