机构地区: 华南农业大学工程学院
出 处: 《华南农业大学学报》 2009年第3期99-102,共4页
摘 要: 以华南地区的3种土壤为材料进行喷灌试验,抽取表征土壤入渗性能的关键特征值,利用神经网络建立了土壤入渗类别识别模型,用试验数据回归建立了灌溉水在各类土壤中的入渗深度预测模型,并对模型进行室内试验验证.结果表明:土壤入渗类别识别模型能对2种检验土进行分类,入渗深度预测模型的预测深度与灌溉水实际入渗深度的误差不超过10%. Sprinkling irrigation experiment were done on three soils in the South China area. Key charac-teristic values of soil infiltration were extracted, and the recognition model of soil infiltrability category was built by using neural network. The infiltration depth predictive model of irrigating water was generated by using experimental data. The verification experiments were done in laboratory. The results showed that the recognition model of soil infiltrability category could sort the two checked soils,The percentage error between the predictive filtration depth and the real depth was less than 10%.