Using AI to predict machine maintenance

XAIPre aims to streamline maintenance in the maritime industry. Current maintenance is often done at regular intervals and with high frequency to reduce the risk of incidents. Because of this, maintenance is often done too early, making it one of the most ineffective industrial activities.

Moreover, existing maintenance systems do not offer insights into what happens between the input and output data. Therefore, it is often unclear why the machine functions improperly, and which actions need to be taken to resolve the issue.

The process could be a lot more efficient and transparent.

The goal is to develop predictive maintenance technology that is both transparent and user-friendly for engineers, as well as insightful regarding the current state of the machine. This project focuses specifically on maintenance technology within the maritime sector.

This project develops a prototype with which engineers can view the current state of a machine from a distance. The sensors in the machine send data to the algorithm about important factors like warmth or friction. The algorithm shows a risk analysis (for example: maintenance is necessary before the ship leaves shore) and the most important factors on which this analysis is based (like: a component warms up quicker than it should). This way, specific and correct action can be taken and unnecessary action can be avoided.

The new system has three big advantages:

Safer: maintenance incidents are the second biggest incident category within the maritime industry. A lack of knowledge is often the cause of these accidents. A clear overview, made by AI and sensor technology shows why the machine functions improperly and which part to work on, increasing understanding and safety. The risk of unsafe working conditions also reduces by preventing unnecessary actions.

More sustainable: Properly functioning machinery uses less power and reduces the need for specialised personnel and parts that have to be transported to remote locations. This results in a smaller CO2-footprint.

More economically viable: Predictive maintenance leads to a reduction of maintenance duration and overall maintenance costs. It also has a positive impact on the lifespan of machines and machine parts.

Because of the collaboration between the industry and research, technology is developed that can be directly implemented in practice.

This project focuses on the maintenance of ship propellors from Heerema Marine Contractors. However, the developed technology can also be applied to other industries and sectors later on. Examples could be: ship design (C-Job Naval Architects), steel production (Tata steel Europe), airplane engines (AirFrance/KLM), and the car-industry (Volkswagen Group and Honda Research Institute Europe).