The EU-funded project, Andromeda, developed a system that integrates smart Industrial Internet of Things (IIoT) devices and Artificial Intelligence (AI) into the very first end-to-end predictive maintenance solution for railway infrastructure. “Railway is the safest and most efficient and sustainable means of motorised transportation,” observes project coordinator and KONUX co-founder and CFO Maximilian Hasler. “Overall, the aim is to help infrastructure managers and other relevant stakeholders improve network capacity and availability, extend asset lifetime and empower employees to make maintenance more efficient.”
First rail predictive maintenance designed for AI solutions
The system continuously monitors and analyses the health of key components and provides actionable recommendations. It ultimately allows for better maintenance planning by helping infrastructure managers to anticipate failures before they happen and to know the optimal time and type of maintenance needed. Specifically, the system improves availability through early warnings of critical conditions and targeted maintenance and helps digitalise railway infrastructure through automated 24/7 condition monitoring. The AI can identify and help prevent failures before they happen. A dashboard provides a round-the-clock overview of all critical switches and their condition. The solution also reduces maintenance costs through failure prediction and effective maintenance actions, and increases asset lifetime through predictive maintenance and quality assurance of conducted maintenance activities. A maintenance quality check identifies which actions were successful and the length of time needed to determine the most economical and sustainable procedures.
Boosting railway system reliability and capacity
High-precision sensors are key to obtaining important insights into the health condition of components. As such, the consortium developed a proprietary, autonomous IIoT device optimised for predictive maintenance applications and extreme environmental conditions. It’s fully certified and meets state-of-the-art security requirements. The device can easily be deployed in the field in under 10 minutes. This timing is key because it doesn’t disrupt regular train traffic. The state-of-the-art architecture makes data management scalable, flexible, reactive and secure. “It ensures that we can get from learning to operations quicker than it was ever possible before,” notes Hasler. “Its modularity allows us to set up new customer environments incredibly quickly and reliably.” The machine learning algorithms are able to generate information that was previously unattainable. “We can now tell our customers how the health condition of their assets will develop in the next 90 days with a hit rate of over 90 %,” explains Hasler. This allows end users to better plan the timing of maintenance measures. As a result, they are in control of their maintenance needs and asset availability. “We can also provide recommendations on the optimal time and type of maintenance needed by measuring and comparing the effectiveness and sustainability of different maintenance actions we see in the field.” By improving the capacity, reliability and cost-efficiency of railways, Andromeda contributes to making them more competitive compared to other less sustainable means of transport. Hasler concludes: “In this way, the project will help countries get closer to achieving their emissions savings goals and making the passenger experience more reliable, convenient and enjoyable.”
Andromeda, maintenance, railway, predictive maintenance, rail, IIoT, railway infrastructure