Improving Predictive Maintenance with Alarm Intelligence
Ampelmann’s offshore gangway systems generate alarm and sensor data that can provide early indications of asset health issues. For service teams, the challenge is to understand which alarm patterns indicate deteriorating asset health, which situations are becoming more critical and where maintenance should be prioritised first.
Ampelmann wanted earlier insight into potential UPS battery issues, so replacement could be prepared before a battery reached a critical state. At the same time, there was a broader need to analyse alarm behaviour across critical components and better distinguish early warnings from more urgent situations.
The goal was to turn raw alarm and sensor data into actionable asset insights that could support maintenance planning, service prioritisation and operational readiness.
Industry
Maritime & Offshore
Tech stack
Azure, Python, TensorFlow, scikit-learn, MLflow
Solution
Yabba Data Doo worked with Ampelmann to analyse alarm data across UPS battery monitoring and other critical asset signals.
By exploring alarm patterns, signal behaviour over time and failure-related indicators, we developed machine learning models that help identify when a battery system or asset component is moving towards a more critical condition. The analysis focused on recognising relevant patterns in operational data and making those insights useful for service teams.
The solution helped distinguish early warning signals from situations that require more immediate attention. This created a practical foundation for smarter fault detection, maintenance prioritisation and future asset monitoring improvements.
Value
The project helped Ampelmann move from reactive alarm handling towards more proactive maintenance decision-making.
By identifying relevant alarm patterns earlier, service teams can better anticipate battery replacement needs, prepare spare parts in time and prioritise maintenance based on urgency and expected operational impact. This helps reduce the risk of avoidable downtime and supports higher system uptime.
The work also created a stronger data foundation for asset management. By turning raw sensor and alarm data into structured insights, Ampelmann gained a clearer view of asset health and a basis for further improvement across critical systems.