利用传统的聚类算法对直觉模糊集进行聚类分析时，存在对异常值敏感、复杂度较高的问题，不适用于大规模直觉模糊数据的聚类。针对上述问题，提出了一种基于密度峰值思想和加权兰氏距离的直觉模糊聚类算法(WIFDPL)，用来提高算法对直觉模糊数据的检测精度，降低算法的复杂度。由于现有直觉模糊距离算子不满足距离度量的定义，提出了一种新的直觉模糊兰氏距离算子，减少了数据的偏移程度，降低了对异常值的敏感程度；由于凝聚型层次聚类算法复杂度较高，采用密度峰值聚类算法对直觉模糊集进行聚类，显著提高了算法的运行效率。实验结果表明，利用改进的直觉模糊兰氏距离提高了聚类精度，且新算法复杂度较低，更适用于大规模直觉模糊集的聚类。

In the clustering analysis of intuitionistic fuzzy sets, the traditional clustering algorithm is sensitive to outliers and has high complexity, so it is not suitable for clustering of large-scale intuitionistic fuzzy data. To solve the above problems, an intuitionistic fuzzy clustering algorithm (WIFDPL) based on density peak and weighted Canberra distance is proposed, which can improve the detection accuracy of intuitionistic fuzzy data and reduce the complexity of the algorithm. Since the existing intuitionistic fuzzy distance operator does not satisfy the definition of distance measure, a new intuitionistic fuzzy Canberra distance operator is proposed, which can reduce the deviation degree of data and reduce the sensitivity to outliers. Due to the high complexity of condensed hierarchical clustering algorithm, density peak clustering algorithm is used to cluster intuitionistic fuzzy sets, which greatly improves the running efficiency of the algorithm. Experimental results show that the clustering accuracy is improved by using the improved intuitionistic fuzzy Canberra distance, and the new algorithm is more suitable for clustering large-scale intuitionistic fuzzy sets with lower complexity.