Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Supervised Time Series Classification for Anomaly Detection in Subsea Engineering

Ergys Çokaj, Halvor Snersrud Gustad, Andrea Leone, Per Thomas Moe, Lasse Moldestad

TL;DR

This work addresses anomaly detection in subsea engineering by framing a time series classification problem on simulated Orcaflex data for intact vs broken wellhead states. It compares dispersion-based preprocessing (STD and covariance transforms) followed by PCA with several classifiers (Logistic Regression, DT, SVM) and a 1D CNN, across multiple noise levels. The key contributions are a systematic evaluation of classic ML methods on dispersion-enhanced features, demonstration that PCA reduces dimensionality effectively, and evidence that CNNs yield the highest accuracy albeit with higher training costs. The findings support data-driven, automated monitoring in offshore operations and suggest future directions toward one-class and unsupervised anomaly detection to handle unobserved crack events.

Abstract

Time series classification is of significant importance in monitoring structural systems. In this work, we investigate the use of supervised machine learning classification algorithms on simulated data based on a physical system with two states: Intact and Broken. We provide a comprehensive discussion of the preprocessing of temporal data, using measures of statistical dispersion and dimension reduction techniques. We present an intuitive baseline method and discuss its efficiency. We conclude with a comparison of the various methods based on different performance metrics, showing the advantage of using machine learning techniques as a tool in decision making.

Supervised Time Series Classification for Anomaly Detection in Subsea Engineering

TL;DR

This work addresses anomaly detection in subsea engineering by framing a time series classification problem on simulated Orcaflex data for intact vs broken wellhead states. It compares dispersion-based preprocessing (STD and covariance transforms) followed by PCA with several classifiers (Logistic Regression, DT, SVM) and a 1D CNN, across multiple noise levels. The key contributions are a systematic evaluation of classic ML methods on dispersion-enhanced features, demonstration that PCA reduces dimensionality effectively, and evidence that CNNs yield the highest accuracy albeit with higher training costs. The findings support data-driven, automated monitoring in offshore operations and suggest future directions toward one-class and unsupervised anomaly detection to handle unobserved crack events.

Abstract

Time series classification is of significant importance in monitoring structural systems. In this work, we investigate the use of supervised machine learning classification algorithms on simulated data based on a physical system with two states: Intact and Broken. We provide a comprehensive discussion of the preprocessing of temporal data, using measures of statistical dispersion and dimension reduction techniques. We present an intuitive baseline method and discuss its efficiency. We conclude with a comparison of the various methods based on different performance metrics, showing the advantage of using machine learning techniques as a tool in decision making.
Paper Structure (20 sections, 21 equations, 23 figures, 8 tables, 1 algorithm)

This paper contains 20 sections, 21 equations, 23 figures, 8 tables, 1 algorithm.

Table of Contents

  1. The data set under consideration
  2. Exploratory data analysis
  3. Standard deviations transformation
  4. Covariance transformation
  5. Principal Component Analysis
  6. Baseline method
  7. Logistic Regression
  8. Experiments
  9. Decision trees
  10. Experiments
  11. Support Vector Machine
  12. Experiments
  13. Convolutional Neural Networks
  14. Experiments
  15. Comparison of methods
  16. ...and 5 more sections

Figures (23)

  • Figure 1: Stack with sensors and corresponding data
  • Figure 2: Two 1-hour simulations from the dataset comparing a broken and intact well under similar conditions. Plots are given for the $x$ and $y$ component of the different physical measurements. The two top rows give the time series while the bottom row shows phase plots.
  • Figure 3: Pair plot showing of the scatter and distribution of data after a standard deviation transform (left). Plot visualizing the transformed data in 3 dimensions (right).
  • Figure 4: Pair plot of the data after using aforementioned covariance transform. For certain combinations the broken and intact cases separate quite well.
  • Figure 5: Ratio each component explains.
  • ...and 18 more figures