An Ensemble-Based Approach to Anomaly Detection in Marine Engine Sensor Streams for Efficient Condition Monitoring and Analysis

Abstract

This study proposes an unsupervised anomaly detection method using sensor streams from the marine engine to detect the anomalous system behavior, which may be a possible sign of system failure. Previous works on marine engine anomaly detection proposed a clustering-based or statistical control chart-based approach that is unstable according to the choice of hyperparameters, or cannot fit well to the high-dimensional dataset. As a remedy to this limitation, this study adopts an ensemble-based approach to anomaly detection. The idea is to train several anomaly detectors with varying hyperparameters in parallel and then combine its result in the anomaly detection phase. Because the anomaly is detected by the combination of different detectors, it is robust to the choice of hyperparameters without loss of accuracy. To demonstrate our methodology, an actual dataset obtained from a 200,000-ton cargo vessel from a Korean shipping company that uses two-stroke diesel engine is analyzed. As a result, anomalies were successfully detected from the high-dimensional and large-scale dataset. After detecting the anomaly, clustering analysis was conducted to the anomalous observation to examine anomaly patterns. By investigating each cluster’s feature distribution, several common patterns of abnormal behavior were successfully visualized. Although we analyzed the data from two-stroke diesel engine, our method can be applied to various types of marine engine.

Health assessment framework of marine engines enabled by digital twins

“…The engine brake power of an individual cylinder is calculated by utilising the in-cylinder pressure, according to equation (10).…”

“…In specific, data-driven approaches have been experiencing significant uptake for marine machinery applications, and include the collection of critical measurements, such as exhaust gas temperatures, lubricating oil temperatures and jacket cooling water temperatures. These can be analysed using Artificial Neural Networks (ANNs), [6][7][8] support vector machines 9 and various unsupervised learning techniques, 10,11 to detect abnormal deviations. More elaborate approaches include condition forecasting and prognostics, such as prediction of the Indicated Mean Effective Pressure (IMEP) changes using Deep Belief Networks (DBNs), 12 prognostics of turbocharger condition using Long-Short Term Memory (LSTM) networks 13 and prediction of the remaining useful life of critical machinery using Convolutional Neural Network (CNNs).…”

Measurement Methods in the Operation of Ships and Offshore Facilities

“…The first article by D. Kim et al is entitled, “An Ensemble-Based Approach to Anomaly Detection in Marine Engine Sensor Streams for Efficient Condition Monitoring and Analysis” [ 4 ]. The authors proposed an unsupervised anomaly detection method using sensor streams from a marine two-stroke diesel engine to detect anomalous system behaviors that may be a sign of system failure.…”

Explainable Anomaly Detection Framework for Maritime Main Engine Sensor Data

“…However, their approach suffers from low performance when the dataset involves high dimensional spaces [ 23 ]. Recently, Kim et al [ 24 ] proposed an ensemble-based method that is scalable to high-dimensional and large-scale sensor data.…”

“…Given an anomaly found, the user cannot easily understand which sensor is most responsible for this anomaly. Previously, Kim et al [ 24 ] tried to explain the anomaly pattern by performing the clustering analysis with anomalous data. However, their method could not quantify the degree of contribution for each sensor on found anomalies.…”