Machine Learning Promises Sustainability in Shipping Sector


  • Uzmar research and development leader Nalan Erol and business development executive Emre Çaylak explain how machine learning is used for safe, sustainable tug operations.
  • Machine learning is a branch of artificial intelligence (AI) and computer science which uses data and algorithms to imitate the way humans learn, gradually improving its accuracy.  
  • In the marine industry, the data comes from sensor measurements and design variables. Machine learning has three main processes – decision, error and optimisation.  

Machine learning promising for sustainability. 

Evaluation of prediction of the model

During the decision process, the algorithm estimates a pattern in the data. An error function then evaluates the prediction of the model. Finally, the model is optimised.  

If the model in the training set fits well to the data points, weights are adjusted to reduce the difference between the known example and the model.  

There are several algorithms for machine learning. Neural networks simulate the way the human brain works, with a huge number of linked processing nodes.  

These networks are good at recognising patterns and play an important role in applications including natural language translation, image recognition, speech recognition and image creation. 

Linear regression is an algorithm used to predict numerical values, based on a linear relationship between different values. For example, the technique could be used to predict tug prices based on historical data for the area. 

Logistic regression is a supervised-learning algorithm that makes predictions for categorical response variables, such as yes/no answers to questions. It can be used for applications such as classifying spam and quality control on a production line.

Clustering algorithms using unsupervised learning can identify patterns in data so that it can be grouped. Computers can help data scientists by identifying differences between data items that humans have overlooked.  

Decision trees can be used to both predict numerical values (regression) and classify data into categories. They use a branching sequence of linked decisions that can be represented with a tree diagram.  

An advantage of decision trees is they are easy to validate and audit, unlike the black box of the neural network.  

In a random forest, the machine learning algorithm predicts a value or category by combining the results from several decision trees. 

Machine learning in shipping

Machine learning in shipping can be grouped into two applications, naval architecture and operational. In naval architecture, it can be used to estimate ship parameters, shaft power and in computational fluid dynamics studies.  

Operationally, machine learning can be used in automatic ship docking, manoeuvring simulation, trajectory prediction, wind load calculation, cargo optimisation, fuel consumption, collision avoidance and maintenance.  

For example, using sensor data via engines, it is possible to determine how efficiently the engines are running.  

The amount of fuel being consumed can be under control, so the emissions it is producing can be seen simultaneously. In another case, it is possible to predict when equipment is likely to fail, enabling condition-based maintenance and reducing costs.   

Training an algorithm on vessel traffic

In addition, by training an algorithm on vessel traffic, weather and historical casualty data, accident candidates can be identified from historic vessel tracks. Ship repair yards can use machine learning for predictive maintenance of ballast pumps.  

While machine learning can be adapted to both the engineering and operational side of the marine industry, there are big challenges such as data measurement reliability.  

The algorithms need to include physics and domain knowledge for better modelling. As machine learning is seen as black-box, reasoning of the trained models is to be improved and the robustness of the models is also questionable.  

To sum up, it seems both data measuring systems and machine learning have a way to go, but this does not mean the industry should wait before implementing them.


Did you subscribe to our daily Newsletter?

It’s Free! Click here to Subscribe

Source: Riviera


This site uses Akismet to reduce spam. Learn how your comment data is processed.