A Maritime Industry Experience for Vessel Operational Anomaly Detection: Utilizing Deep Learning Augmented with Lightweight Interpretable Models

TL;DR

This paper reports an industry experience applying semi-supervised deep learning to vessel anomaly detection on the sensorized ship TUCANA. It couples a LSTM Autoencoder with a vanilla autoencoder for accurate anomaly detection and augments the outputs with lightweight interpretable surrogate models (random forest and a pruned decision tree) plus a t-SNE visualization to aid engineer assessment. On real-world data featuring propeller faults and critical maneuvers, the LSTM AE achieves high precision and recall, while the surrogate DT provides human-readable rules that align with expert reasoning, enhancing transparency and trust. The framework supports edge deployment, improves engineer decision-making, and points to future directions such as adaptive thresholds and federated learning across vessels to enable broader generalization under regulatory considerations.

Abstract

This study presents an industry experience showcasing a vessel operational anomaly detection approach that utilizes semi-supervised deep learning models augmented with lightweight interpretable surrogate models, applied to an industrial sensorized vessel, called TUCANA. We leverage standard and Long Short-Term Memory (LSTM) autoencoders trained on normal operational data and tested with real anomaly-revealing data. We then provide a projection of the inference results on a lower-dimension data map generated by t-distributed stochastic neighbor embedding (t-SNE), which serves as an unsupervised baseline and shows the distribution of the identified anomalies. We also develop lightweight surrogate models using random forest and decision tree to promote transparency and interpretability for the inference results of the deep learning models and assist the engineer with an agile assessment of the flagged anomalies. The approach is empirically evaluated using real data from TUCANA. The empirical results show higher performance of the LSTM autoencoder -- as the anomaly detection module with effective capturing of temporal dependencies in the data -- and demonstrate the practicality of the lightweight surrogate models in providing helpful interpretability, which leads to higher efficiency for the engineer's decision-making.

Figures (9)

• Figure 1: A scheme of data streaming from TUCANA to data storage and visualization platform.
• Figure 2: Overview of the TUCANA onboard installation
• Figure 3: A daily view of the data signals
• Figure 4: ML-driven operational anomaly detection augmented with lightweight interpretable models, for the TUCANA sensorized vessel
• Figure 5: Detected anomalies by the DL anomaly detection models on test data 1.