New Way Now: Talgo keeps trains on track with 24/7 fleet monitoring and AI agents
20Featured in this video: Isaac Casanova Martinez, IT Project Leader at Talgo, and Jose Antonio Marcos, Talgo’s Director of Smart Maintenance Engineering
Executive summary: Talgo, one of the world's foremost train manufacturers, trusts Google Cloud to help keep its high-speed trains at peak performance. Using Google Cloud, including BigQuery, Dataflow, Compute Engine, and Cloud Run, Talgo developed TSMART© — a real-time monitoring platform that makes critical data available to maintenance teams while enabling predictive maintenance to detect potential failures. Sophia, an AI agent powered by Vertex AI, also provides multilingual conversational assistance for more reliable and efficient problem-solving. Now, technicians and engineers can describe issues in natural language and instantly find relevant information based on past maintenance history and troubleshooting manuals from suppliers. Together, TSMART© and Sophia are ushering in a new era of data-driven maintenance, enabling Talgo to hit its target of more than 1.5 million kilometers between failures and boost train reliability by 30%.
Challenge: Talgo must ensure the highest levels of safety, availability, and reliability for trains that reach speeds over 350 kmph, including guaranteeing its operators more than 1.5 million kilometers between failures. To meet these demanding standards, Talgo wanted to shift its maintenance strategy away from traditional reactive maintenance by developing cutting-edge technology that could harness the vast amounts of real-time data it collects — over five gigabytes daily from each train — to monitor trains on the move and optimize maintenance without disrupting service.
Solution: With Google Cloud, Talgo built TSMART© — a real-time monitoring platform that handles and makes available vast amounts of sensor data about everything from voltage to hydraulics to acceleration and temperature. Using BigQuery, Dataflow, Compute Engine, and Cloud Run, Talgo can collect, ingest, store, and process more than 30,000 signals per second while providing the data needed to fuel AI models that help predict failures and problems before they occur. In addition, Talgo also used Vertex AI to create Sophia, an AI agent that provides conversational assistance to maintenance teams. Sophia can process and interpret multiple languages, including English, Russian, and Arabic, allowing technicians and engineers to describe issues in natural language and find the best solutions possible based on past maintenance history and troubleshooting manuals from suppliers.
Results: With TSMART© and Sophia, Talgo is transforming maintenance to deliver unparalleled safety and reliability on the rails. Now, Talgo can monitor its global fleet 24/7, anticipate problems before they occur, and resolve issues faster — delivering a 30% increase in train reliability and helping achieve over 1.5 million kilometers between failures.
By the numbers 📊
→ 1.5 million+ kilometers achieved between failures, hitting a key target for high-speed rail operators
→ 30% increase in train reliability, keeping trains running safely and on schedule
→ 30,000 signals processed per second from each train to fuel predictive maintenance models
→ 24/7 real-time fleet monitoring across its global operations, powered by Talgo’s AI agent Sophia
Key takeaways and highlights from our interview with Isaac Casanova Martinez, IT project leader at Talgo, and Jose Antonio Marcos, Talgo’s director of TSMART© engineering:
→ “We are using Vertex AI to build an intelligent agent, Sophia, to solve the questions of how to repair the trains… We [developed] this tool based on information we have from the past and the troubleshooting manuals from suppliers. We are able to solve any kind of problem — brakes, engines, communication, anything. Sophia is really helping to optimize the process. It's a new way to provide maintenance services and people are very happy, very excited with the solution.”
→ “TSMART© is a train monitoring system that is able to ingest, store, process, and make the data available for the maintenance engineer. We have petabytes of data stored in BigQuery and all the engineers are able to query data and obtain results in seconds. BigQuery is the best solution for us because it is a serverless solution.”
→ “By using TSMART© , we have been able to increase our reliability more than 30%. Also, with a higher level of safety, because we monitor the main safety system of the train in real time, and also by using the AI we are able to monitor the fleet 24/7. TSMART© has been [a] very good solution not just for Talgo but also for operators.”
Google Cloud products used: Vertex AI, BigQuery, Dataflow, Compute Engine
Learn more:
→ Talgo streaming two thousand events a second at three hundred kilometers ahttps://goo.gle/3WfbLFd/3WfbLFd