An Anomaly Detection Algorithm for Spatiotemporal Data Based on Attribute Correlation

In cyber physical systems (CPS), anomaly detection is an important means to ensure the quality of sensory data and the effect of data fusion. However, the challenge of detecting anomalies in data stream has become harder over time due to its large scale, multi-dimension and spatiotemporal features. In this paper, a novel anomaly detection algorithm for spatiotemporal data is proposed. The algorithm firstly uses data mining technology to dig out correlation rules between multidimensional data attributes, and output the strong association attributes set. Then the corresponding specific association rules for data anomaly detection are built based on machine learning method. Experimental results show that the algorithm is superior to other algorithms.