How edge computing revolutionises railway operations
From failure detection to in-service support and usage-based maintenance
Real-time monitoring in Rail
To prevent failures, to provide accurate real-time support to drivers during service, and to perform maintenance closer to design, Railway companies need to be able to access the data on their trains in an easy, safe, fast and flexible way. The old, complex and asynchronous ways of capturing data with USB sticks and diagnostics PCs take time and are not able to deliver the information people need, exactly when they need it. Real-time monitoring equipment on the other hand, simplifies and automates the data access process. And, it enables operators to analyse the state of their assets faster thanks to the real-time data at hand.
However, getting data from your trains in real time is one thing. Obtaining valuable and actionable insights from that data in just a couple of seconds is a different story!
What kind of information do you need to track to provide those actionable insights to your teams? What do you do with all the redundant information that is also sent to your systems or the cloud? How do you turn that vast amount of data into actionable insights?
This is where edge computing and Railnova’s rule engine come in.
What is edge computing and what is Railnova’s rule engine?
Through real-time monitoring operators are able to gather a lot of data. Often it is all sent to data centers or the cloud, even though some of the data is redundant. This results in very high SIM card costs and high server infrastructure costs. (To give you an idea, a fully decoded MVB represents 600GB (!) of data every month.)
You could downsample the data, to limit the amount of data being sent from your assets to your data centers. But, then you’d risk missing important information and events. Some of these events would be missed even if the data was sent e.g. every second. To avoid data gaps, or drowning in a sea of data, (and to lower the costs) it’s more efficient to only transmit the important and interesting data.
A good way to do this is by performing analytics as close to the data source as possible. This is exactly what happens with edge computing. Instead of sending all the raw data, including the redundant data, to data centers or the cloud for analysis, the smart monitoring device already triages and processes the data on the train itself. Afterwards only the relevant alerts and important contextual data (raw values) are sent to data centers.
Edge computing also allows operators to create new counters and to detect very short events, as there’s no downsampling on the train components.
To communicate with the edge device which information it needs to process on the train and send to the cloud, Railnova developed a rule engine. The rule engine is a software component on the Railster, Railnova’s edge device, that applies logic to multiple data sources to detect events and abnormalities.
Through this rule engine, Railnova clients are able to create their own rules and deploy these on the Railster in real time. The next time an event happens that corresponds with that rule, a message goes out to the software platform Railfleet. From there the operator is able to view and analyse the event, and even provide diagnostics tips to the hotline and teams on the ground.
The benefit of such a rule engine is that the operators are in full control and are able to track whatever is relevant to their operations. The rule engine will also allow operators to combine data from multiple data sources, and to compute algorithms on e.g. the MVB stream at MVB sampling rate to add more context to certain events. In addition, operators are able to create their own fault codes or counters without having to buy expensive certified software upgrades from rolling stock OEMs.
Operators can for example create counters to track
- the number of times a train’s pantographs go up,
- how long the pantographs stay up,
- the number of km per pantograph.
They’re then able to use those counters to accurately plan their usage-based maintenance.
How does Railnova perform edge computing?
We already name-dropped “Railfleet” and the “Railster” in previous paragraphs, but haven’t fully explained what these do and how they enable edge computing on rolling stock. So, let’s have a look!
The Railster is Railnova’s edge device which is installed on rolling stock and connected to the MVB and other data sources. The device sends alerts and data at intervals defined by the operator to Railfleet, or to other systems, for further analysis. This enables operators to have a clear real-time overview of the status of their fleet. It also helps the hotline to provide more context when a driver comes across a fault code.
Fault codes and other data from the train are automatically sent to Railfleet, Railnova’s software platform. This platform offers operators and their partners
- Access to raw train and component data for further analysis
- A real-time overview of the status and availability of their rolling stock
- The possibility to maintain a complete view of all open telematic alerts for their entire fleet, and to create new automated alerts
- An easy way to turn alerts, fault codes and other events into work orders
- An efficient way to plan corrective, usage-based and predictive maintenance
- A powerful integration with ERP and maintenance management systems
As mentioned before, Railnova clients are able to set up their own rules so that they can create new fault codes and counters. These rules are created on the Railfleet platform and then deployed in real time on the Railster. For the rules we use the scripting language Lua.
Lua is a lightweight scripting language that supports data-driven programming. As it’s fast and has a small footprint, it can easily be embedded in applications. This makes it the best language for us to create and change rules on rolling stock components and buses, even while the train is in service.
As data needs are often not all known upfront, many iterations are required to get new data and events. The systems you use need to be able to manage these updates easily. Railnova’s rule engine allows operators to build asset intelligence cost-effectively and independently from OEM systems. Contrary to OEM solutions, the rules on the Railster can be changed as frequently as needed using the device management tool Railfleet. No IT interventions or 1M€ software updates needed. This also means that trains don’t have to go into the workshop every time an update needs to happen.
The moment a new rule has been deployed, the Railster will start sending the required data. The raw data will by default be visible (and downloadable) in the telematics raw data and the data inspector graphs on Railfleet. The data can also be turned into an email or dashboard notification, visible to the people on the alert contact list, or with access to the asset in question.
New rules can be added at any time, which enables operators to start with a couple of use cases and to add more along the way, continuously building on their experience.
How can edge computing and rule engines be used in rail?
Railnova has access to the full MVB in real time, so we can track parameters from HVAC, doors, brakes, bogies, PIS, MMI, sensors etc. for our clients. They can then use the rule engine to create new rules and monitor the behavior of those components.
Two of our clients are already doing so on their TRAXX fleet.
The only counter available by default is the total kilometer counter. Deutsche Bahn (DB) uses the rule engine to create more counters to perform usage-based maintenance. This means that they can track counters for other sub-components too. For example: how many times was the train started, how many times did the pantographs go up or down, how much time did it take to lift them, or how many kilometers did the train run in multiple unit. Tracking these counters enables DB to maintain their assets closer to design, and to avoid over- or under-maintaining them. The rule engine makes it easier and quicker to capture and update the counters.
Other clients are using the rule engine to support them in their ERTMS efforts. Thanks to the rule engine they are able to perform millisecond sampling to compute alerts, and to combine data from multiple data sources on the train and its components. This allows them to create their own fault codes, and to provide extra context for those fault codes. Ultimately it helps them to support their hotline, workshops and teams on the ground with timely alerts and diagnostics tips, thereby limiting and even avoiding delays.
Interested to get started with edge computing and Railnova’s rule engine yourself? Get in touch with us, we’re happy to schedule a call or meet with you in person!
Like what you’re reading? Here’s more cool stuff!
