4Tel is working to bring the latest in artificial intelligence technologies to simplify the uptake of condition monitoring.
In a report prepared for Infrastructure Australia ahead of the first Australian Infrastructure Audit, consultants GHD surveyed the maintenance needs of all major categories of Australian infrastructure. When it came to rail, the report found that maintaining Australia’s diverse rail networks was a high priority and in regional rail in particular there was a high likelihood of a coming maintenance gap.
For the regional rail networks, the combination of competition with road freight and existing infrastructure reaching the end of its useful life left much of these networks facing maintenance issues. As the provider and maintainer of train control technology for the Country Regional Network (CRN), Newcastle-based software and hardware engineering firm 4Tel is on the front line of developing innovative technology solutions that provide the ability to bridge the maintenance gap.
General manager of control systems Graham Hjort describes how condition monitoring has been enhanced on the Country Regional Network through application of an Internet of Things (IoT) approach.
“The I/O ports on selected field signalling and telemetry assets are connected to a modem which connects the ports remotely back into a central asset management system called 4Site, which then allows the health of the asset to be interpreted and, if need be, alarms or reports triggered based on the information received from the asset.”
The process also allows changes to be directed back to the field asset by the reverse connection to change selected settings.
“Another way in which condition monitoring has been improved is through improved analysis of information from the field sites,” Hjort continues. “One of the typical functions that 4Site is able to perform is a real time analysis of how long it takes a set of points to move between positions. If the time taken for those points to move and lock into place is above an acceptable threshold, an alarm is raised via 4Site and an appropriate course of action initiated.
By tapping into the existing telemetry, for remote connectivity, 4Tel has been able to remotely control field assets and their reporting without the need for any additional communications hardware. When you start to talk about return on investment, it is minimal outlay, maximum return.”
While this approach to condition monitoring has its benefits, unless maintenance providers use asset condition information as part of their infrastructure maintenance practices, then the benefits may be illusory.
Many physical rail assets are unable to provide an interface for health information, however 4Tel is using emerging technologies to solve this issue. In 2018 4Tel partnered with the University of Pretoria, South Africa, to understand the role that Artificial Intelligence (AI) and Machine Learning (ML) could play in remotely identifying and assessing the health of rail infrastructure. This relationship, along with an existing relationship with the University of Newcastle, NSW, has proven fruitful by providing a platform for researchers to practically apply their work to solving current issues facing one of the largest industries across the globe. With students from these universities, 4Tel is exploring how AI will improve operations for both train operators and rail infrastructure maintainers.
4Tel’s senior artificial intelligence scientist, Dr Aaron Wong is part of the 4Tel Artificial Intelligence Engineering team that includes staff in Australia and internationally. He also continues his work as a conjoint lecturer at the University of Newcastle.
“The use of AI not only can assist in the identification and analysis of defects and faults, but it can also help to reduce cost and risk by allowing the AI to trudge through the data to identify the areas of concern,” said Wong.
Putting these software-driven solutions into practice has also enabled 4Tel to take condition monitoring beyond signalling and cover a broader range of rail infrastructure.
“AI allows us the ability to move beyond track circuits, points, and interlockings for condition monitoring. AI can be applied to rail, ballast, sleeper, and structural defects,” said Wong.
With rail maintenance vehicles and trains travelling across the network, 4Tel is developing a suite of sensors and cameras which are able to easily be fitted to a range of vehicles to provide continuous monitoring of rail condition. The aim of this project is that faults are able to be identified in real time, geo-located and tagged, and then reported back to a maintainer, said Hjort.
“What we are aiming to do here is detect where the fault is or is developing, and if needed, send the maintenance team information about the defect to allow them to conduct their initial assessments before they’ve even left their depot.”
Wong highlighted that ML teaches the AI system the different characteristics of a fault or defect.
“Then the system will be able to utilise that learning in future assessments to identify these faults as they develop over time,” he said.
The introduction of AI into the rail industry in Australia is just beginning with practical applications across a range of environments.
“4Tel’s AI solution allows for multiple inputs into our AI and Machine Learning application. We are able to cater for all the different environments that impact rail operations including in areas of low light such as tunnels, fog, and other challenging spaces including those with high traffic, with the aim of reducing people in the corridor.” said Wong.
“Once the information has been captured through the sensors and/or cameras, the AI processing mines through the data that is collected and then provides detailed assessments to the maintenance provider on the state or the health of the asset,” he said.
AI can significantly shift the rail industry in Australia to more proactive maintenance structure. While this is an example of 4Tel using AI to monitor the health of rail infrastructure, the application of this technology also extends to the above rail operations.
Railway networks and train operations are going to be extensively impacted by AI-based innovation over the current decade and in the future.