Abstract: This paper presents an outlier pattern detection algorithm of time series based on K-Mean Distance Outlier Factor （K-MDOF）.This algorithm uses edge weight to extract the edge point of time series pattern representation，and then this algorithm extracts the four eigenvalue of each sub-pattern，that is，pattern’s length，pattern’s height，pattern’s mean and standard deviation to map time series to feature space，and finally uses K-mean distance outlier factor to detect outlier pattern in this feature space.Detecting time series’s outlier behavior from pattern’s point of view can recuperate the limitation of point outlier detection’s individual behavior，and enhance the efficiency and veracity of outlier detection.Experiments on synthetic and real data show that the definition of pattern outlier is reasonable and this algorithm is efficient to detect outliers in time series.

摘要： 提出了一种基于k-均距异常因子检测时间序列异常模式的算法（K-MDOF）。该算法首先利用边缘权重因子提取时间序列模式表示的边缘点，然后通过提取每一段子模式的四个特征值：模式长度、模式高度、模式均值和标准差将时间序列映射到特征空间，最后利用k-均距异常因子在该特征空间中检测时间序列的异常模式。从模式的角度检测时间序列的异常行为弥补了点异常检测的个体行为局限性，提高了异常检测的效率和准确性，在仿真数据集和真实数据集上的实验结果都证明了在时间序列异常检测中模式异常定义的合理性以及算法的有效性。