基于相关性分析的工业时序数据异常检测

多维时间序列上的异常检测,是时态数据分析的重要研究问题之一.近年来,工业互联网中传感器设备采集并积累了大量工业时间序列数据,这些数据具有模式多样、工况多变的特性,给异常检测方法的效率、效果和可靠性均提出更高要求.序列间相互影响、关联,其隐藏的相关性信息可以用于识别、解释异常问题.基于此,提出一种基于序列相关性分析的多维时间序列异常检测方法.首先对多维时间序列进行分段、标准化计算,得到相关性矩阵,提取量化的相关关系;然后建立了时序相关图模型,通过在时序相关图上的相关性强度划分时间序列团,进行时间序列团内、团间以及单维的异常检测.在真实的工业设备传感器数据集上进行了大量实验,实验结果验证了该方法在高维时序数据的异常检测任务上的有效性.通过对比实验,验证了该方法从性能上优于基于统计和基于机器学习模型的基准算法.该研究通过对高维时序数据相关性知识的挖掘,既节约了计算成本,又实现了对复杂模式的异常数据的精准识别.

Anomaly Detection on Industrial Time Series Based on Correlation Analysis

Anomaly detection on multi-dimensional time series is one important research problem in temporal data analysis. In recent years, large-scale industrial time series data have been collected and accumulated by equipment sensors from Industrial Internet of Things (IIoT). These data show the feature of diversity data patterns and workflows, which requires high performance of anomaly detection methods in efficiency, effectiveness, and reliability. Besides, there exists latent correlation between sequences from different dimensions. The correlation information can be used to identify and explain anomalies in data. Based on this, this paper proposes a correlation analysis based anomaly detection on multi-dimensional time series data. It first computes correlation values among sequences after standardization steps, and a time series correlation graph model is constructed. Time series cliques are constructed according to correlation degree in the time series correlation graph. Anomaly detection is processed within and out of a clique. Experimental results on a real industrial sensor data set show that the proposed method is effective in anomaly detection tasks in high dimensional time series data. Through contrast experiments, the proposed method is verified to have a better performance than both the statistic-based and the machine learning-based baseline methods. Research in this paper achieves reliable correlation knowledge mining between time series, which not only saves time costs, but also identifies abnormal patterns form complex conditions.