The Indian Railways plans to use artificial intelligence to improve maintenance, looking to cut the number of accidents on its 66,090 km network of tracks.

The railways will trial predictive maintenance of signalling infrastructure next year, which is crucial to keep trains stay clear of each other and prevent collisions.

Sensors will collect data on the health of the systems, and the data will be analysed to pick up any potential failures in real-time, allowing them to be fixed before it malfunctions. The system will predict failures of signalling assets, automatically correct them, and provide advice on decisions officials need to make.

The Indian Railways currently manually monitors the status and maintenance of such infrastructure, with problems fixed only after the systems have failed. Such failures are one of the major causes of train accidents in the country.

In 2015-16, there were 106 rail accidents in India, 60% of which were derailments. A government report on the safety of Indian Railways noted that officials were having “difficulty in giving time” to maintain rail infrastructure.

The predictive maintenance system will use “non-intrusive sensors for continuous online monitoring of signals, track circuits, axle counters and their subsystems of interlocking, power supply systems including the voltage and current levels, relays, timers”, a Railway Ministry official told the Economic Times.

This will be trialled in two sections of India’s Western and South Western Railways, and may be expanded to others based on the results.