SPONSORED CONTENT: To meet their promise as one of our most sustainable modes of transportation, railways in many developed countries are going to need a big upgrade, writes Karsten Oberle, Head of the Global Railway Practice at communications group Nokia.

 

Often burdened with old and ageing infrastructure, they have fallen behind especially in adopting digital technologies. They cannot tackle these problems piecemeal; they need to develop a system-wide digital transformation strategy. The Internet of Things and advanced analytics using machine learning should figure prominently in their thinking.

 

As the world’s governments struggle to deal with overcrowded highways and the sustainability issues around climate change, many have turned to railways as a key part of their future transportation strategy. To meet these ambitious goals, however, will demand greater efficiency and a better digital experience to attract more passengers. Digital will play a key role on the operational side with use cases such as train control systems, asset monitoring, video surveillance, predictive maintenance, intelligent rail infrastructure and operations, freight and passenger information systems, safety systems, and cybersecurity.

 

IoT sensor technologies are capable of gathering information on many aspects of rail operations. For example, 80% of track issues resulting in derailments are caused by vertical displacement of the track. IoT sensors installed on trackside assets can detect vertical displacement and temperature changes. They can also monitor the motions of actuators on the trains and catenary tension swings.

 

Sensor output is just one of the many kinds of data that can inform decisions about trackside operations and maintenance. Data from historical sources such as rail traffic, rolling-stock flows, maintenance logs, and planning and control activities are also important. However, analysing this data, much of it streaming in real time, far exceeds the abilities of human operators.

 

This is where today’s advanced analytics applications are able to handle the huge amounts of data that operators cannot. The applications can initially use historical data to develop models and algorithms that predict when safety, performance or other operational parameters are being exceeded. Over time, the predictions of these models can be tested against actual performance. The analytics applications then use real-time, streaming data to adjust their models to make them more accurate, which is what is meant by machine learning. As more and more IoT sensor data is analysed, the models become increasingly accurate.

 

Trackside predictive maintenance is a key area where these data-driven machine-learning systems can improve safety and reduce costs. There are defined inspection cycles and time-limiting maintenance windows, which put a special emphasis on maintenance planning for the use of repair equipment and personnel. Today’s preventive and conditions-based maintenance can be improved by using IoT and analytics technologies.

 

Many assets fail during operations when using time-based maintenance schedules from equipment vendors. Unfortunately, if you shorten the maintenance schedule too much, you risk wasting money refurbishing or replacing assets that are actually in serviceable condition. Nokia, for example, has developed a rail asset lifecycle optimisation application that uses existing conditions-based asset assessment and improves upon it using analytics and machine learning techniques.

 

Correlating data from IoT sensors, environmental information and historical trends, these kinds of solutions optimise the maintenance modelling for each asset by continually refining its ability to forecast failure time. Over time, it can also create value with prescriptive analytics that identify specific actions to optimise operations or automate responses to allow for faster adjustments of schedules and plans.

 

The Nokia solution also provides data on the consequences if premature failure should occur. This provides the operator with a way to assess risk as well. With the Nokia system, operators are given data-driven visualisations in a wide range of intuitive visual formats to provide context for operators. This allows them at a glance to align risk tolerances, business objectives and processes based on their asset management strategies and capital investment planning.

 

These kinds of ‘smart’ systems can also help in the operations of railway stations. There are applications for smart lighting, smart waste, flood detection and platform ice detection. To take smart waste as an example, as with rolling stock maintenance, the issue with fixed schedules is that oftentimes waste bins are overflowing, other times they are nearly empty. The issue is how to get better at predicting when to empty them.

There are a variety of IoT sensor solutions that measure waste levels in bins. This helps to plan waste pickup schedules. However, while these solutions can tell you the current state, they have no ability to predict when they should be emptied. As a result, your schedule will still need a fair bit of leeway built in.

 

Predictive analytics are able to develop models based on the real-time sensor data and historical data. They can also pull information flows from other areas such as train schedules, passenger traffic patterns or predicted ambient temperatures (better to empty the waste bins sooner during heat waves). The result is a constantly improving waste bin emptying schedule as the analytics program gets better and better at prediction.

 

The smart rail station would ideally link all of the data being collected around the station from a variety of similar applications, including CCTV video data, online ticket sales, local event programming (such as a football match or outdoor concert) to be able to better predict station usage and how best to optimise resources, safety and security across all areas of station operations. It not only saves money; it improves the passenger experience.

 

While this may sound quite futuristic, there are existing solutions for most of these use cases today, and many more are coming. There is a danger, however, that railway operators may be tempted to approach each solution separately. The real power of these applications is the ability to integrate as much data as possible, as the simple smart waste application illustrates very well. This requires operators to take a system-wide approach with a strategy for digital communications, IoT management, and data collection, storage and processing. This system-wide approach is why communications system vendors, such as Nokia, are developing solutions in this space, offering rail operators an integrated platform from which to launch their full digital transformation.