ann-产量评法在耕地地力评价中的应用及养分状态分析——以辽宁省彰武县为例.pdf

摘 要 本文在模糊评价的基础上建立了人工神经网络模型，选取了4 大类、14 个指标 进行评价，得出了彰武县的耕地地力等级分布,并且对耕层土壤的主要养分指标进行 了分析。 （1）利用评价模型得出彰武县的耕地主要以三级（中等水平）地力为主。三等 地的耕地面积占全部耕地面积的40.41%。二等地和四等地的分布面积相差不大，分别 为22.58%和28.52%。一等地和五等地的耕地面积分别占全部耕地面积的比例为4.08% 和4.41%。 (2)彰武县的耕地中碱解氮的总体含量处于缺乏水平的耕地面积占总面积的 49.24%；速效磷的含量在中等以上水平的耕地面积占总面积的89.64%；速效钾处于四 级较缺乏水平的耕地面积占总面积的55.05%。 （3）有效铜的含量处于高水平的耕地面积占总面积的90.6%；有效锌的含量处于 中等水平的耕地面积占总面积的54.54%；有效铁的含量处于中等水平的耕地面积占总 面积的86.83%；有效锰的含量处于高水平的耕地面积占总面积的44.14%。 （4）通过两种评价方法对彰武县耕地地力评价比较研究结果表明，地力等级完 全相同以及等级相差一级的评价单元数目占评价单元总数的87.5%，可见两种方法的 评价结果等级差异较小，表明人工神经网络模型评价方法同样能达到较高的准确度， 具有客观性。 关键词：耕地地力评价；人工神经网络；养分状态 Evaluation model of cultivated land fertility using artificial neural network and its application and analysis of nutrient status —A case in Zhangwu county of Liaoning provice Abstract Artificial neural network model based on fuzzy evaluation in this paper, selected the four categories and 14 indicators to evaluate, obtained the cultivated land fertility distribution of Zhangwu county and analysis the Nutrients in topsoil. (1) The evaluation model derived the farmland to three (medium level) fertilith-based.The area of third-class accounted for 40.41% of all arable land area.The difference of area between second places and four is small, they are 22.58% and 28.52% . The area of first-class and fifth-class accounted for 4.08% and 4.41% of all arable land area. (2)The overall content of available nitrogen in the lack of level in Zhangwu county is 49.24%,the available phosphorous content in the medium and above the level of arable land is 89.64% of the total area,the content of potassium in the four lacking level arable land is 55.05% of the