BP神经网络和SVM模型对施加生物炭土壤水分预测的适用性

发布时间：2018-03-03 00:04

本文选题：土壤水分　切入点：生物炭　出处：《水土保持研究》2017年03期 　论文类型：期刊论文

【摘要】：生物炭作为土壤改良剂对半干旱区土壤水分有良好的吸持作用,为确定施加生物炭对土壤水分预测模型适用性的影响,依托黄土高原半干旱区固原生态站开展了小区定位试验。向土壤中施加不同种类及比例的生物炭,定期监测土壤水分含量;考虑土壤含水量的非线性特征以及生物炭对土壤水分的影响,选取BP神经网络和SVM支持向量机两种模型,建立施加生物炭土壤水分预测模型。计算预测值,并与实测值对比,分析相对误差;利用RMSE、MRE、MAE和R2评估BP神经网络和SVM模型的精度。结果表明;BP神经网络预测值的平均相对误差为3.78%,最大误差为13.14%;SVM模型的平均相对误差为0.56%,最大误差为2.42%。SVM模型的RMSE、MRE、MAE值(分别为0.34~0.17,0.07,0.56~1.27)均小于BP神经网络的(分别为1.04~1.16,0.47~0.68,3.78~4.57),且决定系数R2值SVM模型(0.96~0.99)大于BP神经网络(0.56~0.64)。BP神经网络和SVM模型均能很好地预测施加生物炭的土壤水分,但SVM模型预测结果更加稳定,精度较高,更适于施加生物炭土壤水分的预测。该研究可为半干旱地区生物炭还田土壤水分的预测及管理提供理论依据。
[Abstract]:Biochar, as a soil modifier, has a good absorption effect on soil moisture in semi-arid area. In order to determine the effect of biochar on the applicability of soil moisture prediction model, Based on the ecological station of Guyuan in the semi-arid region of the Loess Plateau, the plot location test was carried out. Different kinds and proportions of biochar were applied to the soil, and the soil moisture content was monitored regularly. Considering the nonlinear characteristics of soil water content and the effect of biochar on soil moisture, two models, BP neural network and SVM support vector machine, are selected to establish the prediction model of soil moisture with biological carbon. The predicted values are calculated and compared with the measured values. Analysis of relative error; The accuracy of BP neural network and SVM model is evaluated by RMSE MREMAE and R2. The results show that the average relative error of BP neural network prediction value is 3.78, the maximum error is 13.14 and the average relative error of SVM model is 0.56 and the maximum error is 2.42.SVM model RMSEMREMAE (. The values of 0.34 ~ 0.17 ~ 0.07 ~ (0.56) ~ (1.27) are smaller than those of BP neural network (1.04 ~ (1.16) ~ 0.47 ~ (0.68) ~ (0.68) ~ (0.78) ~ (3.78) ~ (4.57)) respectively, and the R ~ (2) value of SVM model is 0.96 ~ 0.99), which is larger than that of BP neural network (0.56 ~ 0.64 ~ 0.64). BP neural network and SVM model can predict soil moisture of biochar well. However, the prediction results of SVM model are more stable and accurate, so it is more suitable for the prediction of soil moisture in biochar. This study can provide a theoretical basis for the prediction and management of soil moisture in semi-arid areas.
【作者单位】：西北农林科技大学资源环境学院;陕西省农业厅;西北农林科技大学理学院;中国科学院水利部水土保持研究所黄土高原土壤侵蚀与旱地农业国家重点实验室;重庆邮电大学计算机科学与技术学院;
【基金】：国家自然科学基金“生物炭对黄土高原不同质地土壤水文过程影响及机理的定位研究”(41571225)
【分类号】：S152.7

【相似文献】