Efficient on-line anomaly detection for ship systems in operation

作者：

Highlights：

• Anomaly detection with signal reconstruction and residuals analysis.

• Reduced computation time, due to use of training data clustering.

• Applied on 14 imbalanced datasets, including recent data from a marine diesel engine in operation.

• Regional credibility estimation used in the residuals analysis.

摘要

•Anomaly detection with signal reconstruction and residuals analysis.•Reduced computation time, due to use of training data clustering.•Applied on 14 imbalanced datasets, including recent data from a marine diesel engine in operation.•Regional credibility estimation used in the residuals analysis.

论文关键词：Anomaly detection,Condition monitoring,Maritime industry,Auto Associative Kernel Regression (AAKR),Cluster based AAKR,Sequential Probability Ratio Test (SPRT)

论文评审过程：Received 6 December 2017, Revised 20 December 2018, Accepted 21 December 2018, Available online 22 December 2018, Version of Record 4 January 2019.

论文官网地址：https://doi.org/10.1016/j.eswa.2018.12.040