基于ANFIS的坡地水土流失预测

The Prediction of Soil and Water Loss on Slope Based on ANFIS

针对水土流失预测的多因素及非线性等特征,引入自适应模糊神经推理系统(Adaptive network-based fuzzy inference system,ANFIS),对自然降雨条件下不同生态措施的水土流失规律进行了预测。结合MATLAB模糊推理工具箱,以试验站2001—2005年的51场降雨为训练样本进行学习训练,以2008年观测的10场降雨为检验样本进行验证,结果表明:4个处理(百喜草全园覆盖、百喜草覆盖果园、清耕果园和裸露对照)的径流预测的平均相对误差的绝对值分别为17.04%、18.89%、15.74%和16.61%,平均17.07%;泥沙的预测平均误差的绝对值为20.87%、17.71%、19.28%和16.71%,平均为18.57%;从误差分析可以看出模型具有较高的精度和稳定性,表明ANFIS模型可以有效的描述自然降雨条件下不同生态措施坡地的水土流失特征,而且还可以描述和处理模糊信息。ANFIS模型不但操作简单,计算量小,而且在物理参数较少的情况下也可以取得较高的精度,为自然降雨条件下坡地水土流失预测的进一步研究提供了新的思路。

Based on the principle of momentum theorem and water balance the basic equation for describing the runoff generated on a slope under the condition of natural rainfall with different rainfall intensities, rainfall momen- tum and influence of ecological measures taken into account is deduced. The equation can be numerically solved by applying the Preissmann format. The feasibility and precision of the model is verified by field experimental data. Results show that the mean relative error of the prediction under 3 treatments are 14.97%, 13.59% and 15.15%. The model is quite high in accuracy and stability, and serves as useful tool in further research on prediction of runoff of nature rainfall on slopes under different ecological measures.