冻融坡面土壤剥蚀率不同预测模型比较

发布时间：2018-11-12 08:51

【摘要】：为探究水蚀因子对冻融坡面土壤剥蚀率的影响,采用2个坡度(10°,15°)、4个流量(4.5,6.5,8.5,10.5L/min)和4个起始解冻深度(2,5,10,15cm),模拟野外径流冲刷试验。采用BP神经网络方法和逐步回归分析法,分析土壤剥蚀率和流量、坡度、起始解冻深度、流速、水流剪切力、水流功率与单位水流功率7个水蚀因子关系。结果表明:通过BP神经网络连接权关系分析水蚀因子对冻融坡面土壤剥蚀率影响顺序为水流功率单位水流功率起始解冻深度水流剪切力流量流速坡度。BP神经网络模型的土壤剥蚀率预测平均误差为2.848%(R~2=0.954);逐步回归模型的土壤剥蚀率预测平均误差4.820%(R~2=0.925);基于单一水蚀因子(水流功率)模型的土壤剥蚀率预测平均误差5.298%(R~2=0.867)。基于BP神经网络的土壤剥蚀率预测效果最好,为春季解冻时期冻融坡面不同起始解冻深度条件下土壤侵蚀预报模型的建立提供了新思路。
[Abstract]:In order to investigate the effect of water erosion factor on soil erosion rate on frozen and thawed slope, two slopes (10 掳, 15 掳), four flow rates (4.56.58.5L / min) and four initial thawing depths (21015cm) were used to simulate the field runoff scour test. BP neural network method and stepwise regression analysis were used to analyze the relationship between erosion rate and flow rate, slope, initial thawing depth, flow velocity, flow shear force, water power and water erosion factor per unit flow power. The results show that the influence of water erosion factor on soil erosion rate on frozen and thawed slope surface is analyzed by BP neural network connection weight relationship. The order of soil erosion rate is the gradient of flow velocity and flow velocity of flow shear force and flow rate of thawing depth of water flow per unit power of flow power and BP neural network. The average prediction error of soil erosion rate of the complex model was 2.848% (RP20.954). The average prediction error of soil erosion rate of stepwise regression model was 4.820% (Ru 20.925), and that of soil erosion rate based on single water erosion factor (water power) model was 5.298% (RP20.867). The prediction effect of soil erosion rate based on BP neural network is the best, which provides a new idea for the establishment of soil erosion prediction model under different initial thawing depth of freezing and thawing slope in spring thawing period.
【作者单位】：沈阳农业大学水利学院;
【基金】：国家自然科学基金项目(41471225) 辽宁省农业领域青年科技创新人才培养计划(2014054)
【分类号】：S157.1

【相似文献】