A few weeks ago, the Federal Railroad Administration asked for input from the rail industry and the public about the role automation and artificial intelligence (AI) can play in rail. It is a sign that FRA recognizes the rail industry will need to leverage technology to safely and efficiently transport more freight and passengers by rail in the coming decades across the U.S.
Railroads have looked to technology before to remain the most efficient and safest mode of transportation. Positive Train Control (PTC), once fully implemented, will be one of the most significant advances in railroad technology and safety. PTC will prevent many incidents caused by human error: overspeed conditions, collisions, entering an established work zone or misrouting due to a misaligned switch.
What will be the next technological breakthrough? Artificial Intelligence. No other technology today holds the promise AI does to further improve safety and increase efficiency. So-called “PTC 2.0” is expected to advance the technology to a level well beyond a safety overlay system and provide real business benefits that help the bottom line. AI will play a large role in that.
Just like other industries, rail has vast troves of data taken from sensors on locomotives, wayside equipment and other systems that can be converted into intelligence. When railroads understand these data insights and can easily put them to work, they can predict and prevent failures, delivering huge benefits to the bottom line, to customers and to safety. AI is the key to making that happen.
North American Class I railroads we work with are already tapping the power of AI and realizing the benefits. Through that experience, we have seen AI keep trains in revenue service from breaking down. Across hundreds of locomotives being monitored by AI, more than 200 mission-critical failures have been prevented during the past year alone. By eliminating breakdowns, railroads have been able to maintain and improve rail network velocity, reduce delayed shipments to customers and increase the safety of communities through which railroads operate.
By keeping trains moving and preventing critical locomotive failures, network velocity is improved, reducing the number of delayed shipments to customers.
Railroads also use AI to perform virtual load tests, instead of manual tests, which are still the norm. These tests are usually conducted twice per year to verify that a locomotive is producing the rated horsepower needed for propulsion and dynamic braking.
By capturing the thousands of signals from locomotives and the engine control system, and then using AI and data science to analyze it, electricians can view a locomotive’s performance as if it were going through a manual load test. When virtual load tests are used, railroads can annually save 255 gallons of diesel fuel for each locomotive tested. Multiply that by the thousands of Class I locomotives in service, and the numbers add up fast.
The virtual tests have other benefits: They allow mechanics time to gather the parts and equipment needed to repair a locomotive before it arrives, reducing the amount of time workers are in harm’s way.
AI technology is here, and it’s in use today. We encourage the FRA and the railroads to learn as much as they can about it , to fully realize its safety and efficiency potential.
Joe Becker is Managing Director for Rail at Uptake, an industrial AI company headquartered in the nation’s rail hub of Chicago.