Skip to main content
European Commission logo
TRIMIS

Predictive Maintenance for railway switches. Smart sensor networks on a machine learning analytics platform

Project

Andromeda - Predictive Maintenance for railway switches. Smart sensor networks on a machine learning analytics platform


Funding origin:
European
European Union
STRIA Roadmaps:
Transport infrastructure (INF)
Transport infrastructure
Transport mode:
Rail
Rail
Transport sectors:
Passenger transport
Passenger transport
Freight transport
Freight transport
Duration:
Start date: 01/12/2017,
End date: 01/11/2019

Status: Finished
Funding details:
Total cost:
€2 155 000
EU Contribution:
€1 508 500

Overview

Objectives:

92,000,000 minutes, or 180 years. These are the (potentially avoidable) train delays caused by problems due to rail switches – every year, in Europe alone.

The rail switch is a critical component of the rail-infrastructure. It is responsible for actively changing the route of trains, and must therefore endure higher levels of stress and mechanical fatigue than other elements of the rail infrastructure. Failures at track switches can result in expensive delays, or even dangerous derailments. In fact, track switches account for 19% of the delays, and have been linked to 110 derailments since 2010.

It is therefore unacceptable that most of the monitoring of critical switches is solely dependent on manual inspections, since they are prone to measurement error and lack of global oversight. The outcome is generally an inaccurate estimation of the 'State-of-Health' of each switch and re-active – not preventive – maintenance, with financial losses due to delays or additional investment in infrastructure maintenance.

European spending on railway switches is currently over 4B€ (30B€ worldwide). Furthermore, passenger and freight volumes are expected to increase by 34% and 40% by 2030 respectively, which means shorter infrastructure access time for maintenance and monitoring.

KONUX has developed the ANDROMEDA system, which promises to revolutionize the maintenance procedures for rail infrastructure. The system pulls data from sensors deployed permanently at the rail switch, which means no more expensive and manual data acquisition. Through self-calibrating smart sensors and machine learning analytics, the result is a holistic overview of the whole rail switch network, with current (and predicted) 'State-of-Health' for each switch. With a total of 600,000 switches in Europe and 2,510,000 switches worldwide, the ANDROMEDA project represents a 12.4B€ business opportunity.

Contribute! Submit your project

Do you wish to submit a project or a programme? Head over to the Contribute page, login and follow the process!

Submit