What if you could monitor major infrastructure – including dams and bridges – using machine learning? No humans required.
Typically, infrastructure is monitored using video footage taken from helicopters or drones, and then staff or contractors watch that footage to identify problems. It’s a very time consuming process, and there is a chance of human error – people failing to spot faults.
Late last year, researchers at the University of Surrey and King’s College London published a paper in the journal, Structural Health Monitoring, detailing how they created an Artificial Intelligence (AI) system – called SHMnet – to analyse and assess damage to bolt connections in metallic structures.
They tasked SHMnet with accurately identifying the changes to connection bolts on a steel frame in 10 different damage scenarios. After training SHMnet using four repeated datasets, and then running tests, it was able to produce a 100% identification record. That is, it identified every visible fault.
One of the researchers, Dr Ying Wang, Assistant Professor at the University of Surrey explained, “The performance of our neural network suggests that SHMnet could be incredibly useful to structural engineers, governments and other organisations tasked with monitoring the integrity of bridges, towers, dams and other metal structures.”
In fact, similar technology is already being developed commercially and used by utilities companies. Birmingham-based start-up, Keen AI, works with National Grid to speed up identification of defects or faults in their towers, fittings and conductors.
Keen AI’s KAI-Vision platform uses deep learning to identify components automatically and present these to an operator for review. It has meant National Grid has reduced the time taken to review overhead line footage by 66%.
The benefits are not only related to time and cost savings. For an engineer, the reviewing process is laborious and has significant bottlenecks. And the quality of the work is dependent on the individual engineer reviewing the footage. AI takes human error out of the equation.
The input from Keen AI’s initial work with National Grid is being used to train a machine learning algorithm to allow the model to automatically identify defects, saving further time. It has also enabled the organisation to remove the backlog of footage waiting to be processed, which reduces the risk of towers and lines failing – consequently reducing the risk of homes and businesses experiencing loss of power.
This technology could be deployed in a number of different industries, and there is the potential for it to be used to monitor dams, bridges and other structures. This could help pick up defects and problems ahead of time, avoiding the issues experienced at the Whaley Bridge Dam in 2019.
It’s an interesting time to be working in our industry, as new technologies mean that changes to working practices can happen rapidly. And if this enables us to keep our infrastructure in better condition and less likely to fail, it’s a positive development.
Meanwhile, AI has not yet learnt how to be a structural engineer, so if you need assistance with the structural elements of an upcoming project, please do get in touch.