The railway and public transport industry is like an ocean liner that always moves very little by little. For once, the sector has been surprised by the rapid adoption of artificial intelligence (AI), applied both to manufacturing processes in industrial plants and at the traveler level by public transport operators.
The evolution is such that in maintenance companies the typical image of the operator with a yellow vest and helmet next to the tracks is being replaced by a guy sitting in the office controlling everything from a computer. In most cases the data was already obtained with existing technology, but there was no way to take advantage of it.
Anticipation of incidents through patterns that detect anomalies
Alstom has created an international working group in its offices in Madrid, headed by Victor Martín, with a combination of engineers from the sector along with data scientists, computer scientists and machine learning specialists. “We teach the AI the patterns of when there may be a problem on the road and, based on the large amount of data received, the system is able to detect friction at a point that anticipates a degradation of the infrastructure and alert the team. maintenance”, exemplifies Martín. In this way, they can anticipate the resolution of all types of incidents and resolve them before they fail definitively, thus saving the impact on the passenger service that it had before, when it was going to be repaired once the breakdown had occurred because before It was undetectable.
The amount of information generated by trains and the infrastructure managed by large multinationals such as Alstom was unimaginable until now. For Martín, “the key is the patterns created from the data”, thus being able to know the health of the motor of a needle change based on data such as the seconds it takes to perform a specific operation. If you start taking more time than usual, the warning goes off.
They know this well at Transports Metropolitans de Barcelona (TMB), which promoted the creation of Smart Motors, a company specialized in predictive maintenance that did similar things long before AI was talked about. Its detection system has been applied to needle changes for a decade.
More recently, they have equipped one train on each line with a system that reads all the data on the road infrastructure and allows any possible incident to be detected, marking the exact point where staff should go to make the most of the nighttime maintenance period, according to explains the head of digitalization of the TMB metro, Ignasi Oliver. The incident reports from the trains themselves also help them to anticipate possible incidents that affect the comfort of passengers, such as a failure in the air conditioning.
At Siemens, among other things, they apply it to improve the safety of level crossings. “The images captured with a camera are combined with a system for detecting people, vehicles and large and small objects to improve safety and interrupt traffic if necessary,” explains Katja Elschner, head of technology at Siemens Mobility, highlighting that “ The reliability of AI is greater than that of humans as long as it has been given accurate data and is used for specific things.”
The images captured by cameras in trains and stations allow demand to be managed in detail
Hitachi, for its part, applies a very similar technology to respond to the needs of public transport operators using cameras. “AI allows us to better plan future scenarios and anticipate problems,” says Ed Brown, from the Japanese manufacturer. For example, they have a system that detects travelers who are carrying a bicycle.
Internally, it allows us to know the demand of people who are committed to intermodality and to better plan the spaces designed for bikes inside the wagons. At the same time, the same technology installed next to the validating machines can also be used with direct implications for the traveler by limiting their access if bikes are prohibited from entering during rush hour due to lack of space.
It is in this area where there is more room for exploration, with direct effects for travelers. “Machine learning from AI makes it possible to predict user flows within convoys and at stations and thus make appropriate decisions to improve the service,” exemplified Henri Poupart-Lafarge, CEO of Alstom, during the inaugural session of Innotrans, the largest exhibition in the railway sector, held in Berlin a few weeks ago.
Dedicating the first round table to this issue was already a declaration of intentions in a scenario that in previous years focused on talking about financing or decarbonization. Also at that same table was the CEO of the Spanish CAF, Javier Martínez Ojinaga, who praised the open possibilities but also warned that “there is an abuse of the concept, as happened with big data a few years ago,” and urged “ look for specific cases that can really take advantage of AI and think about the corresponding business model.”
“AI introduces as big a change to the sector as the creation of the internet did,” responded Michael Peter, CEO of Siemens Mobility. Of course, the open spirit of the web at the time is what Peter claims for this new era in the railway, “with an open architecture that can benefit all” manufacturers and that, in turn, has an impact on users.
Read also
#railway #sector #train
