中文摘要：

为了优化同异反（Identical-Discrepant-Contrary,简称IDC）灰色相关分析中同异反趋势划分方法,提高同异反趋势划分结果的精度,文章在分析两种传统划分方法存在不足的基础上,对其进行改进,提出了均分迭代划分法和回归系数比值划分法,并结合土壤中有机质含量和砷含量相关性的实例,对改进后的两种方法进行数值模拟.结果表明：改进后的两种划分方法得到结果的可靠度均较高,采用均分迭代划分法得到结果的可靠度为70%,采用回归系数比值划分法得到结果的可靠度为55%,略低于前者,这是因为土壤有机质含量中存在＂异常＂数据,对回归系数影响较大,降低了回归系数比值划分结果的精度.

英文摘要：

For the sake of optimizing the Identical-Discrepant-Contrary（IDC） trend division method in the IDC grey correlation analysis, and improving the accuracy of the IDC trend classification results, this paper based on analyzing the deficiencies of the two kinds of traditional classification methods, improved it, and put forward the average-division iterative method and regression coefficients ration method. Combining with the example of relativity between the content of organic matter and the content of As in soil, this paper carried numerical simulation on the two improved methods. The results show that： The results getting from two classification methods that were improved both have a high reliability. The reliability of the result getting from average-division iterative method is 70%, and from regression coefficients ration method is 55%, slightly lower than the former. Because the content of the organic matter in the soil has ＂abnormal＂ data, it has a great influence on the coefficient of regression, and reduces the accuracy of division result getting from regression coefficients ration method.