Global
Predictive maintenance of vessel’s power generator​
Predictive Maintenance of Vessel’s Power Generator    Case Study Energy
Overview

An MLOps predictive maintenance solution to reduce unnecessary periodic interven​

Background

The project consisted in developing a ML solution to avoid redundant maintenance interventions and replacements on a crucial vessel’s electric power generator.​

Solution

Approach:

The system lives both on-shore (on cloud) and off-shore (on the vessel). The on-shore part of the solution is devoted to regularly read fresh generator’s data synced from the vessel to the Azure cloud, and together with CMMS events automatically retrains a machine learning model every month. Once trained, the new model is then moved off-shore to perform online predictions every 12 hours. The model advices if an intervention is needed and explains what generator sensors are driving the prediction in order to support the maintenance intervention. The results are visualized through dashboards: on-shore implemented with Power BI, and off-shore with ChartJS.

Results

Results Achieved:

MLOps best practices have been implemented integrating Azure Machine Learning and Azure Devops services.​

The off-shore routines are managed by using docker containers orchestrated with Kubernetes​

Ready to collaborate and create value?

Let’s get in touch and see a new vision of the future!

Enter your details, one of our experts will get in touch soon