Express trains Full speed ahead thanks to AI-based maintenance

Related Vendors

FAULHABER Drive Systems

Taiwan Machine Tool & Accessory Builders' Association (TMBA)

China’s use of artificial intelligence (AI) in maintaining its extensive high-speed rail network is quite impressive. The network, which spans 45,000 kilometers and supports trains traveling at speeds of up to 350 kilometers per hour, is now reportedly “in better condition than when it was built

*Henrik Bork, long-time China correspondent for German "Süddeutsche Zeitung" and General Manager of AsiaWaypoint, Beijing , PR China. 

The 45,000-kilometer (approx 28,000 miles) network of the Chinese Gaotie express trains, which travel at speeds of up to 350 kilometers per hour (217 mph), is proactively maintained using artificial intelligence (AI) and is now "in better condition than when it was built," writes the lead engineer in a scientific journal.

Last year, thanks to proactive maintenance with the help of AI, for the first time none of the express train routes in operation received a single warning that necessitated a speed reduction.

An AI control center in Beijing processes huge amounts of data collected by sensors, cameras and track workers. It can generate warnings for urgently needed repairs "within 40 minutes", report Niu Daoan, Chief Engineer of the China State Railway Group, and his colleagues in an article for the scientific journal "China Railway". Last year, thanks to proactive maintenance using AI, for the first time "not a single warning requiring a speed reduction was received on any of the express train lines in operation", the report states.

The accuracy of reports for necessary maintenance work has reached 95 percent and the "number of minor rail faults has decreased by 80 percent compared to the previous year", write the Chinese railroad engineers. "This helps the teams on site to carry out new inspections and repairs as quickly as possible," writes Niu, who heads the "Infrastructure Inspection Center" of the China State Railway Group.

Maintenance AI to combat skills shortage

The sheer extent of the Chinese high-speed rail network, now the largest in the world, makes "scientific support" for maintenance a necessity, the article states. China is constantly expanding the network. All cities with more than 500,000 inhabitants, of which there are more than 240 in China, are to be connected. With a top speed of 350 kilometers per hour (217 mph), Chinese express trains are currently the fastest in the world. From next year, trains will be allowed to travel at speeds of up to 450 km/h (280 mph) on some routes.

For such a large rail network, which requires intensive maintenance at such high speeds for safety reasons, it is almost impossible to find enough track attendants, even in densely populated China. Average incomes are rising and there are fewer and fewer people who want to do this job. For this reason, an idea that was developed in Germany and Switzerland just over ten years ago has been taken up—to make extensive use of artificial intelligence to improve maintenance efficiency, the Chinese railroad engineers write.

How the AI was trained

It is well known that AI needs large data sets to be useful. While the rail network in Germany and Switzerland, especially for high-speed trains, is quite small, this specific problem was solvable in China. They worked hard for several years to initially collect 195 terabytes of data, Niu and his team write. 57 different types of data from 23 categories now flow together in the control center of the Chinese railroad, from "dynamic waveform values" from sensors mounted on the wheels, to vibration values of the rails, to manually collected data from the track attendants and meteorological records.

Several dozen algorithms were then integrated in order to be able to comprehensively analyze the "seemingly chaotic data sets", the report continues. Thanks to "big data", very fast and even predictive warnings about real and impending disruptions are now possible. The railroad engineers report that it has even been possible to effectively reduce rail fluctuations caused by strong winds on long railroad bridges spanning wide valleys.

The challenge of maintaining high-speed rail networks

Scientific support for proactive safety alerts and "precise infrastructure maintenance of high-speed rail networks" is an important milestone for transportation, the paper says. When the first Chinese high-speed train set off 15 years ago, critics wrote that the maintenance of such a network would be almost impossible to manage.

If you add up the length of all the high-speed rail tracks in China, they are already longer than the equator. Because the network is relatively young, there are currently far fewer derailments and other dangerous incidents in the People's Republic than in the ageing rail network in the USA, for example. With the help of AI, China's maintenance experts now want to ensure that this remains the case in the future.