针对综合评价过程中隶属函数建立存在主观性和随机性以及部分系统缺乏指标阈值的问题，引入模糊聚类的思想，建立基于FCM理论的评价模型。当指标阈值存在时，通过阈值确定FCM的最佳聚类中心，得到隶属度矩阵；不存在时，通过AP聚类确定FCM的初始聚类中心，改善传统算法对聚类中心初值选取的随机性；再利用改进的FCM算法对指标数据进行分级评价，得到隶属度矩阵并建立指标阈值，最后进行综合评价分析；并将该模型应用于四川某水域的水质评价中。结果表明，该模型评价结果处于单因子评价和传统模糊综合评价结果之间，其相关系数均在0.7以上，说明该模型结果具有合理性，并且能克服因单因子评价模型仅强调最坏指标和传统模糊综合评价中人为选择隶属函数而导致评价结果具有片面性和主观性的不足。

In the process of comprehensive evaluation, there are subjectivity and randomness in the establishment of membership function, and some systems lack of index threshold. The idea of fuzzy clustering and establishes the evaluation model based on FCM theory is introduced. When the index threshold exists, the best clustering center of FCM is determined by the threshold, and the membership matrix is obtained. When it does not exist, the initial clustering center of FCM is determined by AP clustering, which improves the randomness of the initial value selection of clustering center in traditional algorithm. Then, the improved FCM algorithm is used to grade the index data, and the membership matrix is obtained and the index threshold is established. Finally, the comprehensive evaluation and analysis are carried out. The model is applied to the water quality assessment of a river of Sichuan. The results show that the evaluation results of the model are between the results of single factor evaluation and traditional fuzzy comprehensive evaluation, and the correlation coefficients are all above 0.7, which shows that the results of the model are reasonable and can overcome the shortcomings of one sidedness and subjectivity caused by the single factor evaluation model only emphasizing the worst indicators and the artificial selection of membership function in traditional fuzzy comprehensive evaluation.