SmartShip – Digital twins for intelligent ships and ship fleets
Sea rescue is an essential part of the maritime world. Their task of search and rescue at sea is becoming increasingly demanding as the expanding shipping traffic makes rescue operations more complex and difficult. Technological advances such as automation and digital data processing can help the crew to cope better with these challenges. Machine learning (ML) methods and the use of artificial intelligence (AI) have the potential to support ship crews and significantly improve the outcome of rescue operations. Such techniques can enhance crew capabilities to optimize ship maintenance and search at sea. For this project, it is necessary to equip ships with appropriate sensors and network them appropriately.
SmartShip I (Completed: 01.09.2020 – 31.12.2024)
For this purpose, ships are to be equipped with new sensors. These include camera systems that record the surroundings and provide image data for evaluation. IT/AI systems are to be converted for testing purposes and linked to a digital twin. The aim is to develop a digital twin that monitors the machines on board and makes predictions based on the sensor data. It is hoped that these predictions will extend the service life of the machines by allowing early intervention and maintenance work to be carried out. A digital twin is particularly suitable for this use case, as it is able to process heterogeneous data (AIS, weather, images, etc.) in equal measure and use it for predictions. Forecasts are based on the deviation of sensor data of the actual states with sensor data of the target states.
An additional digital twin is to be developed for entire fleets of ships in order to analyze their behavior. This information can be used to better plan, coordinate, and carry out future missions and search and rescue operations. The knowledge gained can then be used to optimize the fleets. Combined with anomaly detection systems with real-time data acquisition, costs can be reduced by minimizing docking times and operational resources.
SmartShip II (Ongoing: 01.01.2025 – today)
In the past, digital twins have been developed, ML algorithms for anomaly detection have been operated and a prototype of a camera system for object recognition has also been developed. Research is now focusing on standardizing the data architectures and algorithms in order to expand the scope of the models and prepare them for efficient AI use. The digital twins developed so far are to be standardized for different ship classes and adapted so that they are directly AI-compatible. In this way, we can ensure that they can be easily combined with other AI technologies. One of these technologies is Prognostics & Health Management (PHM), which is concerned with maintaining the health of the system.
Large Language Models (LLMs) can represent and reproduce complex specialist domains. If such a component is combined with a PHM system, the crew can benefit from this human-machine interaction. In order to continuously enrich the LLM with expert knowledge, a linguistic interface is to be developed via which the ship’s crew can communicate with the PHM.
Poor weather conditions are a well-known challenge in sea rescue, making the search for shipwrecked and missing people more difficult. Camera systems based only on the visible spectrum are often inadequate, as they are also severely impaired by fog and rain. This is why SmartShip is now aiming to combine several camera systems (e.g. thermal cameras) and radar signals in order to be able to evaluate several sources. The findings and data collected should help to improve the accuracy of object detection.
SmartShip is funded by the German Federal Ministry of Defense (BMVg) as part of the DTEC funding program.
Letzte Änderung: 7. February 2025