| 本文已被：浏览 1088次 下载 739次 |
码上扫一扫！ |
|
|
| 基于云遗传BP神经网络的黄淮海旱作区土壤有机质预测精度分析 |
|
徐清风1,2,3,于茹月1,2,3,勾宇轩1,2,3,赵云泽1,2,3,李勇1,2,3,黄元仿1,2,3*
|
|
1.中国农业大学 土地科学与技术学院, 北京 100193;2.自然资源部农用地质量与监控重点实验室, 北京 100135;3.农业农村部华北耕地保育重点实验室, 北京 100193
|
|
| 摘要: |
| 为探究提高土壤有机质预测精度的方法，以黄淮海旱作区为研究对象，分别运用云遗传BP神经网络、BP神经网络和GABP神经网络三种方法比较不同土层的土壤有机质预测精度。结果表明:1)不同土层土壤有机质值的数据分布与正态分布相比具有不同程度的向右偏移，不同土层土壤有机质均属于中等程度变异；2)不同土层土壤有机质的半方差函数最优拟合模型均为指数模型，不同土层土壤有机质的结构因素与随机因素对空间变异的影响大小基本一致，空间自相关性较弱；3)结合云模型与遗传算法的BP神经网络对0~10、10~20、20~30 cm土层土壤有机质的预测精度均得到了一定提升，而对30~40 cm土层土壤有机质的预测精度则提升不明显，可能是由于30~40 cm土层土壤有机质变异系数超过了一定范围所造成。研究结果可为提高土壤有机质的预测精度提供参考，并为进一步调整耕地管理措施及提高土壤质量水平提供依据。 |
| 关键词: 云遗传模型 BP神经网络 土壤有机质 黄淮海旱作区 |
| DOI：10.11841/j.issn.1007-4333.2021.04.15 |
| 分类号: |
| 基金项目:国家重点研发计划(2016YFD0300801) |
|
| Prediction precision analysis of soil organic matter based on cloud genetic BP neural network in Huang-Huai-Hai dry farming area |
|
XU Qingfeng1,2,3,YU Ruyue1,2,3,GOU Yuxuan1,2,3,ZHAO Yunze1,2,3,LI Yong1,2,3,HUANG Yuanfang1,2,3*
|
|
1.College of Land Science and Technology, China Agricultural University, Beijing 100193, China;2.Key Laboratory of Agricultural Land Quality of Ministry of Natural Resources, Beijing 100035, China;3.Key Laboratory of Arable Land Conservation(North China)of Ministry of Agriculture and Rural Affairs, Beijing 100193, China
|
| Abstract: |
| In order to improve the prediction precision of soil organic matter, the dry farming area of Huang-Huai-Hai was studied, three methods, cloud genetic BP neural network, BP neural network and GABP neural network were used to compare the prediction accuracy of soil organic matter in different soil layers. The results showed that: 1)The data distribution of soil organic matter values in different soil layers displayed different degrees of right deviation compared with the normal distribution. The peak was steeper and the two tails were more widely distributed, which belong to the medium degree of variation. 2)All the semi-variance function models of soil organic matter in different soil layers were exponential models. The influence of structural factors and random factors on the spatial variability of soil organic matter in different soil layers in the study area were basically the same, and the spatial distribution tends to be fragmented. 3)The BP neural network combined with cloud model and genetic algorithm improved the prediction precision of soil organic matter in 0-10, 10-20, 20-30 cm soil layers, but not in 30-40 cm soil layer, which might be caused by the variation coefficient of soil organic matter in 30-40 cm soil layer exceeding a certain range. The results can provide a reference for improving the prediction accuracy of soil organic matter, and a basis for further adjustment of cultivated land management measures and improvement of soil quality. |
| Key words: cloud genetic model BP neural network soil organic matter Huang-Huai-Hai dry farming area |
