Real-time AI enabled rail track inspection and analysis [RAPPID]

点击翻译按钮获取中文摘要

Current inspection of rail track defects utilises Network Rail's four Ultrasonic Testing Units (UTUs) that traverse the UK network, 64,000 miles of track, in 750 shifts per year. With a limitation of 30 miles per hour for rail track inspection, UTUs cannot meet the high demand and increased capacity of customers.Every day, 4.8 million people travel by train in Britain. Around 200,000 tonnes of freight and goods are transported by rail in that same time frame, supporting businesses and consumers, productivity, and economic growth whilst taking thousands of lorries off the road, and helping in the reduction of greenhouse gasses.A risk-free network of rail tracks across the UK is pivotal to Network Rail's long-term planning process strategy and its vision for running a safe, reliable, efficient and growing railway, in Control Period 6 and beyond. Undiscovered rail track defects lead to asset failure, unscheduled maintenance, timetable delays, accidents, and fatalities. Train delays cost passengers 3.6 million hours in 2016, whilst over £72M was claimed by passengers from operators for service disruptions in 2016/17\. With the growing demand on rail transport by passengers, there is need for commercial solutions that offers high-speed (i.e. above 60 miles per hour) high resolution, rail track inspection, and data analysis in real-time. A commercial solution with the capacity to enable UK network-wide coverage.This RAPPID project seeks to address the challenges that the UK rail network faces regarding rapid high-speed high-resolution identification of rail track defects, data collation and analysis, enabling real-time predictive analysis, and predictive maintenance of rail tracks across the UK network and globally.The RAPPID project is based on the novel use of Virtual Source Aperture non-destruction testing techniques in combination with artificial intelligence and deep-learning methodologies that enable real time data processing and analysis of rail track data derived via use of next generation phased-array ultrasonic testing hardware.