Learning states for detecting anomalies in time series
Salvador, Stan Weidner
MetadataShow full item record
The normal operation of a device can be characterized in different operational states. To identify these states, we introduce a segmentation algorithm called Gecko that can determine a reasonable number of segments using our proposed L method. We then use the RIPPER classification algorithm to describe these states in logical rules. Finally, transitional logic between the states is added to create a finite state automation. Multiple time series data may be used for training, by merging several time series into a single representative time series using dynamic time warping. Our empirical results, on data obtained from the NASA shuttle program, indicate that the Gecko segmentation algorithm is comparable to a human expert in identifying states, and our L method performs better than the existing permutation tests method when determining the number of segments to return in segmentation algorithms. Empirical results have also shown that our overall system can track normal behavior and detect anomalies. Additionally, if multiple time series are used for training, the model will generalize to cover unseen normal variations and time series that are "between" the time series used for training.