Session: 06-17-05 AI Technology for Ocean Engineering V

Paper Number: 127763

127763 - Application of Data Analytics and Machine Learning in Analysing Result of Numerical Ship Manoeuvring Study 

Introduction: We have carried out an extensive simulation survey of a vessel steering towards and passing through an enclosed ship tunnel in diverse weather conditions, using SINTEF Ocean's time domain vessel simulator VeSim (see [1]). While a companion paper is being submitted to OMAE2024 describing the proposed techniques for carrying on the numerical studies (see [2]), the current paper focuses on application of tools form machine learning and data analytics to analyse the results from the above mentioned simulation study.

In recent years, advancements in AI, data analytics, and machine learning have revolutionized how entities can analyse and interpret vast amounts of data. In this paper, we explore the application of these technologies to predict the success or failure of a safe ship passage through an enclosed tunnel based on thousands of numerical simulations. By leveraging AI algorithms and data analytics techniques, we can uncover hidden patterns and correlations within the simulation parameters, leading to more accurate predictions and informed decision-making.

The paper will be presented with following structure.

Data Collection and Preprocessing: This section will provide a brief description of the simulation study. It will further explore how can we characterise each time domain simulation run with a set of computed parameters from the simulation study. 

Exploratory Data Analysis (EDA): This section will present an EDA to gain insights into the relationships between simulation parameters and the operation's success/failure. Visualizations and statistical analyses can reveal initial patterns, outliers, and potential variables of interest.

Feature Selection: Use AI-driven feature selection techniques such as Recursive Feature Elimination or feature importance from tree-based models to identify the most influential parameters. This step ensures that the model focuses on the essential variables, improving both accuracy and efficiency.

Building Predictive Models: This section will present how supervised machine learning algorithms like Random Forest, Support Vector Machines, or Gradient Boosting can be used to build predictive models. The models will be trained on a portion of the data and validate their performance using another subset (or cross-validation techniques) to prevent overfitting.

Model Evaluation and Fine-Tuning: We will further evaluate the models using appropriate metrics (accuracy, precision, recall, F1-score) and fine-tune them by adjusting hyperparameters to further optimize the models for the best performance.

Conclusion: This section will summarize how incorporating AI, data analytics, and machine learning techniques into the analysis of simulation parameters can significantly enhances the predictive capabilities of determining the success or failure of the safe ship passage. By leveraging these technologies, expert team can make data-driven decisions, optimize operational processes, and improve overall efficiency.

 

References.

[1] https://www.sintef.no/en/software/vesim/

[2] Andrew Riss, et al "Ship manoeuvring study of a vessel transiting a ship tunnel" submitted for review to OMAE2024

 

Presenting Author: Vahid Hassani SINTEF Ocean

Presenting Author Biography: Vahid Hassani is a Senior Researcher at SINTEF Ocean, Norway.

Authors:

Vahid Hassani SINTEF Ocean
Andrew Ross SINTEF Ocean
Leiv Aspelund Multiconsult Norway ASA