Digital transformation in the rail sector is no longer a future promise, but a palpable reality. Artificial intelligence has become a key ally in improving efficiency, safety and passenger experience. In this blog, we will explore how at Exceltic we are applying AI in real rail signalling projects, with measurable results in cost savings and time reduction.<\/p>\n\n\n\n
The use of railway signalling systems is essential to ensure safety, traffic control and the correct operation of infrastructures. However, the design, validation and maintenance of these systems requires an enormous technical effort, which has traditionally been manual, time-consuming and error-prone. This is where artificial intelligence comes into play. <\/p>\n\n\n\n
At Exceltic, we have been exploring for years how artificial intelligence can optimise different phases of a signalling system's lifecycle: from initial engineering to final validation. Thanks to sensorisation, big data and machine learning, it is already possible to automate tasks such as plan comparison, logic and video validation, and even predictive maintenance analysis. <\/p>\n\n\n\n
One of the most obvious uses of AI in the railway sector is the preventive and predictive maintenance<\/strong>. Thanks to the sensorisation of trains, bogies, brakes or rails, intelligent systems can anticipate component wear before a breakdown occurs. This not only saves costs, but also prevents unplanned outages. <\/p>\n\n\n\n For example, if the sensors detect that the brake discs are in good condition, the system can recommend extending their replacement cycle, thus avoiding unnecessary maintenance. At a more advanced level, the smart lanes<\/strong> can analyse vibrations, noises or temperatures to anticipate buckling or cracking. <\/p>\n\n\n\n According to a study by MarketsandMarkets<\/em>The use of AI in rail maintenance is growing by more than 10 % per year. The most advanced solutions already combine physical sensors with: <\/p>\n\n\n\n Predictive maintenance systems are being powered not only by physical data, but also by visual processing and contextual intelligence, extending the life of key components and minimising outages. <\/p>\n\n\n\n In the part of operation and traffic<\/strong>AI makes it possible to manage the flow of trains dynamically. Analysis of images or sensors in stations helps to predict passenger flow and adjust train frequency in real time. It also improves the traveller experience<\/strong>The range of services on offer ranges from virtual assistants that provide information on estimated time of arrival, to personalised tourist recommendations along the way. <\/p>\n\n\n\n In the webinar we share three specific applications where we are already applying artificial intelligence in real signage projects: .<\/p>\n\n\n\n Hundreds of drawings are revised on a railway project. This task, usually manual, can involve dozens of hours of work by draughtsmen. With AI, we can automatically compare versions of plans<\/strong>We can detect changes, inconsistencies or errors, and generate reports in seconds. This frees up technician time for more valuable tasks. <\/p>\n\n\n\n Based on validated drawings, we have developed a system that automatically detects and extracts signalling elements<\/strong> (signals, circuits, needles, etc.) and represents them on the control centre synoptic. This drastically reduces control software development time, speeding up the engineering process. <\/p>\n\n\n\n Our most innovative solution is a visual validation tool<\/strong> which analyses screenshots from the railway control centre. From the images, the system detects visible items, reads labels, recognises colours and changes of state, and automatically checks if the results are correct<\/strong> during the tests. <\/p>\n\n\n\n This system is also able to detect errors such as: <\/p>\n\n\n\n Over the last year, the use of:<\/p>\n\n\n\n For example, when simulating a track change, the system analyses whether the colour of the section changes appropriately, whether the signals behave as they should and whether the alarm or rejection messages appear correctly. This is done both at initial time (T0) and after a manoeuvre, detecting errors and generating complete validation reports. <\/p>\n\n\n\n In addition to validating the logic and the image, we have also developed a comparison between the original plans (PDF)<\/strong> and the generated video synoptic. This allows for the detection of errors such as misplaced signals, incorrect circuit names or missing elements, ensuring consistency between design and implementation. <\/p>\n\n\n\n This dual approach - logical and visual - allows for a validation 100% automatic<\/strong>The process is designed to minimise human error and reduce the effort required at each stage. <\/p>\n\n\n\n In a real case involving a railway station with more than 2,600 movements to be validated<\/strong>we applied this technology with outstanding results. While manual validation required an enormous effort in each loop (test cycle), our automated system allowed: <\/p>\n\n\n\n In addition, the development effort is focused on the loop 1<\/strong>while the following ones (loop 2, loop 3...) require only a review of the detected non-conformities, further reducing the cost. <\/p>\n\n\n\n In parallel to technological developments, international bodies have started to standardise the use of AI in critical environments such as railway signalling. <\/p>\n\n\n\n The European initiative Shift2Rail<\/em> and other regulatory frameworks are driving:<\/p>\n\n\n\n In railway signalling projects, commissioning deadlines are often immovable. Any delays in earlier phases must be compensated for during testing. This is why, shortening validation cycles<\/strong> without sacrificing quality is key to meeting schedules and improving project profitability, especially in highly competitive tenders. <\/p>\n\n\n\n Our AI solution does not replace the human team<\/strong>It enhances it, allowing it to focus on strategic tasks rather than repetitive validations. <\/p>\n\n\n\n Exceltic is a Spanish private equity consultancy 100%<\/strong>founded in 2005. We have more than 650 professionals and offices in Madrid, Barcelona, Bilbao, Valencia and Valladolid. We work with more than 150 clients in 45 countries, and in 2024 we will have more than 150 clients in 45 countries. 38 million euros annual turnover<\/strong>32% of which 32% corresponds to international activity. <\/p>\n\n\n\n In the railway sector we offer services in all phases of the life cycle: signalling, RAMS, maintenance, telecommunications, cyber security, BIM, embedded systems, software, FPGA hardware, and more. <\/p>\n\n\n\n Our railway AI division<\/strong> leads innovation within the group, with solutions that are already proving their worth in real projects. <\/p>\n\n\n\n If you would like to know more about our developments, or if you think we could help you optimise your railway processes with artificial intelligence, please do not hesitate to contact us. At Exceltic we are committed to efficiency, safety and innovation in the transport of the future. <\/p>\n\n\n\n
Intelligent technology applied to traffic control and traveller experience<\/strong><\/h2>\n\n\n\n
Exceltic case studies: validation automation and agile development<\/strong> <\/h3>\n\n\n\n
1. Automated plan comparison<\/strong> <\/h3>\n\n\n\n
2. Automatic generation of the synoptic and control centre elements<\/strong> <\/h3>\n\n\n\n
3. Machine vision-based video validation<\/strong><\/h3>\n\n\n\n
\n
\n
Matching between shots and videography: quality assurance<\/strong> <\/h2>\n\n\n\n
Real impact: up to 42% time savings<\/strong> <\/h2>\n\n\n\n
\n
Global trends: interoperable and certifiable AI<\/strong> <\/h2>\n\n\n\n
\n
Why is this innovation key?<\/strong> <\/h2>\n\n\n\n
Who are we at Exceltic?<\/strong> <\/h3>\n\n\n\n
Interested to know more?<\/strong> <\/h3>\n\n\n\n