基于遗传算法的BP网络在小流域侵蚀量预测中的应用  

Application of Predicting Soil Erosion in the Small Watershed Based on BP Network Model by Genetic Algorithm

作　　者：段军彪[1] 景旭[1] 上官周平[2] 

机构地区：[1]西北农林科技大学信息工程学院,陕西杨凌712100 [2]西北农林科技大学水土保持研究所,陕西杨凌712100

出　　处：《西北农业学报》2008年第2期317-320,共4页Acta Agriculturae Boreali-occidentalis Sinica

基　　金：中国科学院西部行动计划项目(KZCX2-XB2-05)

摘　　要：人工神经网络可用于流域水土流失的预测。针对BP神经网络收敛速度慢及容易陷入局部最优解的缺点,设计了基于遗传算法(GA)的优化BP神经网络。利用遗传算法特有优势,为BP网络的初始权值和阈值搜索全局最优解空间,经过BP算法迭代训练,进行预测。依据黄土高原沟壑区杨家沟小流域多年径流与泥沙的实测数据,对创建的侵蚀量模型进行训练和预测,取得了较高的预测精度和较快的收敛速度。Artificial neural network （ANN） has been applied in predicting soil and water loss of a watershed. Aimed at the shortcomings of the BP neural network, such as slow convergence, easily getting local optimums, a combined BP network using genetic algorithm is proposed in this paper. The original weights and bias of BP network are defined in global optimization based on genetic algorithm, then the neural network for predicting was trained based on back-propagation algorithm. According the measured runoff and sediment data in Yangjiagou watershed in gully region of Loess Plateau, the erosion prediction model with high convergence rate and precision was built and trained.

关 键 词：BP网络 遗传算法 侵蚀 预测 小流域 

分 类 号：S157[农业科学—土壤学]