Exploration of Anomalies in Cyclic Multivariate Industrial Time Series Data for Condition Monitoring

Industrial product testing is frequently performed in cycles, resulting in cycle-dependent test data. Monitoring the condition of products under test involves analysis of large and complex test data sets. Main tasks are to detect anomalies and dependencies between observation variables, which appears to be challenging to engineers. In this paper, we present a flexible and extendable visual analytics approach for anomaly detection focusing on cycle-depended data. It is based on a glyph representation to visualize anomaly scores of cycles with respect to interactively selected reference data. Our approach is built on a design study in collaboration with an industrial engineering corporation, and is demonstrated on real data from engines tested on automotive testbeds. Based on findings from evaluation results, we provide a discussion and an outlook for future work.