Predictive maintenance

Data analytics, machine learning and optimization.


This project involved using sensor data and machine learning to create a system for predictive maintenance, as well as automatic anomaly detection and advanced visualization.


  • Development of autonomous vessels and real-time predictive maintenance based on sensor data.

  • Deployment onshore and offshore, simultaneous support for data sharing/fleet learning and autonomous offline predictive capability.

Sonat Service

  • Architecture and lead for the implementation of machine learning-driven predictive maintenance.

  • Combined Azure cloud resources with on-vessel edge computing for secure real-time sensor data streams, support for intermittent connection, and automated anomaly detection.

  • Provided a prototype for advanced visualization of world wide fleet location, current alerts, weather and real time location video.

The Results


  • Successful deployment of a predictive solution offshore and onshore.
  • Replaces labor intensive (costly, slow) condition monitoring, representing considerable savings on high value assets.
  • Enables scaling to total fleet of thousands of vessels as well as third party partners.
  • Product sold globally.

Predictive solution

scaling to total fleet of vessels

Product sold globally

First we established the required system architecture.

Then we trained, deployed and evaluated models for

automated anomaly detection and fault labeling for common problems.

Sonat AI

Data analytics, machine learning and optimization.


Kjell Værøy Ljøstad

CEO Oslo
+47 952 67 129

Sonat Consulting Oslo AS
Karl Johansgt. 25
0159 Oslo


Jonny Klemetsen

CEO Bergen
+47 294 56 245

Sonat Consulting Bergen AS


C. Sundts gate 17-19

5008 Bergen


Sonat Consulting Bergen AS

Postboks 234 Sentrum

5804 Bergen