Du, S.Ma, X.Li, X.Wu, M.Cao, W.Pedrycz, W.2024-05-192024-05-1920239798350303759https://doi.org/10.1109/CAC59555.2023.10450590https://hdl.handle.net/20.500.12713/41862023 China Automation Congress, CAC 2023 -- 17 November 2023 through 19 November 2023 -- -- 198194Anomaly detection is essential to ensure the safety of industrial processes. This paper presents an anomaly detection approach based on the probability density estimation and principle of justifiable granularity. First, time series data are transformed into a two-dimensional information granule by the principle of justifiable granularity. Then, the test statistic is constructed, and the probability density and cumulative distribution functions of the test statistic are calculated. Next, the confidence level determines the test threshold. Finally, the time series data of a key parameter in the sintering process is used as a case study. The experimental result demonstrates that the proposed approach can detect abnormal time series data effectively, providing an accurate and effective solution for detecting time series anomalies in industrial processes. © 2023 IEEE.eninfo:eu-repo/semantics/closedAccessAnomaly DetectionPrinciple Of Justifiable GranularityProbability Density EstimationTime SeriesConference Object136413672-s2.0-8518931592110.1109/CAC59555.2023.10450590N/A