Rumo invests in AI to improve broken rail detection and reduce accidents

Rumo, the largest railway operator in Brazil, has innovation as one of its business premises and is treading a path of leadership in technological transformations in the sector. In this sense, the company has just implemented the second phase of the Broken Rail Detection (DTQ) project, which consists of integrating an Artificial Intelligence system capable of increasing the levels of assertiveness in identifying problems on the road from 95% to 98%.

The DTQ project was developed by Rumo’s Research and Development (R&D) area, with the first phase starting in 2018, and represents an unprecedented technology in Brazil. In its first phase, the initiative consisted of installing a device at various points on the railway network, and, with or without a train on the track, the conditions of the tracks were passed on to the Network Monitoring Center (NOC). When an anomaly was identified, the drivers of all trains were informed in real-time, eliminating the risk of derailment.

Now, with the implementation of Artificial Intelligence features, the company has optimized the detection of broken rails. “In August and September of this year, without the AI, 96 broken rails would have been detected. With the addition of the model, we had an additional detection of 6 fractures, totaling 102 cases and representing a gain of 6.25% in accuracy,” says Rodrigo de Souza, Railway Technology Manager at Rumo. After the tests were concluded, 60% of the DTQs installed throughout Rumo’s network received the update and entered the assisted operation phase.

Another benefit of the system is the reduction of false alarms for broken rails. “The integration of the system reduced the notifications that caused the train to stop or the need for speed restriction,” explains Souza. “In the tests, 14 false alarms were identified that would require inspections by the track maintenance team. With this new model, we increased the efficiency of the operation and reduced the rate of cases of this type by 50%”, he highlights.

Sharing is caring!