中文摘要：

针对标准集对分析存在的同异反评语细化问题,利用原创联系度可展性对其进行改进,并建立了地表水水质评价模型,可以得到评价样本对于各水质评价等级的优劣趋同程度。并与模糊评价相结合应用于山西省地表水质评价,并引入差异系数来优化评价指标体系,利用熵权系数法和超标倍数赋权法求得指标的综合权重。将评价结果与灰色聚类法、综合指标法、模糊神经网络法和投影寻踪法的评价结果进行比较,结果表明：改进后的的集对分析评价结果更贴近实际情况,为水环境分析提供了一种更为科学、合理的评价和决策方法,具有良好的应用前景。

英文摘要：

This paper proposes an improved set pair analysis（SPA） method using malleability of original relation degree to solve the problem of subtle differentiation between uniform,different and reverse in standard SPA.A water quality evaluation model was developed with this method for evaluation of the excellent and inferior tendency degree of the samples in the corresponding rank.It introduces a discrepancy coefficient and an entropy method to optimize the index system and adopts a superstandard multiple method to determine the synthetic weight of index.Combined with a fuzzy method,it was applied to evaluation of the surface water quality in Shanxi province.A comparison of the SPA method with gray classification method,synthesis index method,fuzzy neural networks and projection pursuit method,shows that its evaluations are closer to the real situation.It is a better method for evaluation and decision-making of water environment and has good potential application.