摘要： 提出了一种简单高效的多维离散时间序列符号化方法，该方法用模糊自适应共振理论(Fuzzy ART)对多维时间序列数据进行聚类，实现多维时间序列数据的符号化问题。同时，通过属性相关性预处理分析，过滤掉聚类中不相关或弱相关的属性，保证了聚类算法的准确性，将提出的算法应用于多维交通流数据的符号化，效果很好。

关键词: 模糊自适应共振理论；多维时间序列；符号化；聚类

Abstract: This paper proposes an efficient method for the symbolization of multi-dimensional discrete time series. This method applies fuzzy adaptive resonance theory (ART) to select the subset of the multi-dimensional time series data, which are the clustering of the time series data and symbolizing them. At the same time, the irrelated or weakly related attributes are filtrated by the study of attributes interrelated analysis that ensures the clustering’s veracity.

Key words: Fuzzy adaptive resonance theory; Multi-dimensional time series; Symbolization; Clustering