摘要
针对矿区传统的土地复垦适宜性评价方法,运用模糊数学的隶属函数法,对决策树的ID3算法加以改进,建立基于改进决策树的矿区土地复垦适宜性评价方法。以平禹二矿为研究区,将改进的评价模型运用到塌陷区土地复垦适宜性评价中,选取7个评价指标作为测试属性,提取100个已确定复垦方向的评价单元作为样本,运用改进的ID3算法,构建研究区土地复垦适宜性评价决策树,然后运用生成的决策树分类规则,对各评价单元按其属性进行分类,最终确定各评价单元的复垦方向,并将分类结果与模糊综合评价方法得到的评价结果进行对比分析。结果表明:基于改进决策树模型的矿区土地复垦适宜性评价方法,不依赖于经验知识,依靠样本的自主学习,制定分类规则,得到的分类结果较为客观合理,而且决策树分类规则确定后,可以直接应用于同一地区的其他待评价区,且分类过程简单明了,便于操作,避免传统评价方法中较为繁琐和复杂的计算过程,提高矿区土地复垦适宜性评价的效率,为矿区土地复垦适宜性评价工作提供一种新的思路。
[Background]The essence of evaluating land reclamation suitability in mining areas is a type of classifying the land reclamation for the future according to certain evaluation rules（ or criteria）. Thus,it is feasible and reasonable to apply classification rules with decision tree induction method in suitability evaluation of land reclamation in mining areas. [Methods] In this paper,we applied the subordinate function method of the fuzzy mathematics to improve ID3 algorithm（ the Iterative Dichotomiser 3） of the decision tree,established a model for evaluating land reclamation suitability,and then applied this improved model in Pingyu Second Mining Area. Selecting 7 evaluation indexes as the attributes of measurement,and extracting 100 units which directions of land reclamation have been determined as the samples,we established the decision tree for land reclamation suitability evaluation with the improved ID3 algorithm. Then,we applied the established classification rules of the decision tree to classify them according to their attributes,to evaluate various units and to determine the reclamation directions forvarious evaluated units. Further,to verify the accuracy of evaluation results,we compared the most frequently used limit condition method with the proposed method. [Results]The statistics result showed that there were 365. 87 hm2 suitable for arable land,98. 83 hm2 suitable for woodland and 97. 98 hm2 suitable for meadow by using improved decision tree model in the studied area. However,there were334. 95 hm2 suitable for arable land,106. 26 hm2 suitable for woodland and 121. 47 hm2 suitable for meadow by using limit condition method in the studied area. The 92. 50 % of evaluation results of the proposed method were consistent with those by limit condition method. In inconsistent results,the grades of evaluation result by limit condition method were lower than those by decision tree model. After comparing the two methods,the results revealed that the limiting condition method was more suitable for rec
出处
《中国水土保持科学》
CSCD
北大核心
2016年第6期35-43,共9页
Science of Soil and Water Conservation
基金
国家自然科学基金"生态约束条件下的区域土地利用结构与空间格局优化研究"(41301617)
中国博士后科学基金"基于遥感的矿区土地复垦质量评价与监测研究"(2016M590679)
中国博士后科学基金"煤矿区土地生态质量评价及变化机理研究"(2015M580629)
河南省高等学校重点科研项目"典型煤矿区土地生态质量时空变化规律与机理研究"(17A420001
16A420003)
关键词
ID3算法
改进决策树
矿区
土地复垦
适宜性评价
ID3 algorithm
improved decision tree
mining areas
land Reclamation
suitability evaluation
