Daily Mail 13 July 2017
System can analyze plasma to help design safe reactors that harness the power of stars on Earth
- Machine learning can be used to spot processes that are causally linked
- This can help to predict what events cause complications in plasma behaviour
- Researchers say it could be used to avoid disruptions that lead to energy loss
Researchers have tapped into artificial intelligence to help overcome some of fusion energy’s greatest challenges.
By feeding a machine-learning program data from past experiments, it can reveal links between processes that cause complications in the plasma’s behaviour
This could help to avoid such disruptions, which lead to rapid loss of stored thermal and magnetic energy, and can even threaten the machine itself.
According to the researchers, this approach could be used to analyze the behaviour of plasma inside a tokamak. And, with the information generated by the machine-learning program, they can then build a system that monitors for the warning signs
Fusion involves placing hydrogen atoms under high heat and pressure until they fuse into helium atoms.
When deuterium and tritium nuclei – which can be found in hydrogen – fuse, they form a helium nucleus, a neutron and a lot of energy.
This is down by heating the fuel to temperatures in excess of 150 million°C, forming a hot plasma.
Strong magnetic fields are used to keep the plasma away from the walls so that it doesn’t cool down and lost it energy potential.
These are produced by superconducting coils surrounding the vessel, and by an electrical current driven through the plasma.
For energy production. plasma has to be confined for a sufficiently long period for fusion to occur.
The team from the US Department of Energy’s Princeton Plasma Physics Laboratory (PPPL) found that machine learning can be used to spot events that may bring on off-normal plasma behaviour inside the tokamak.
And, it doesn’t need to reveal how these events may be linked – all it has to do is note the link itself.
‘When you use machine learning, you consider the models produced by the computer program to be black boxes – you put something into it, and then get something out, but don’t always know how the output is related to what you put in,’ said graduate student Matthew Parsons.
‘In this paper, I make that black box a little more transparent.’
According to the researchers, this approach could be used to analyze the behaviour of plasma.
And, with the information generated by the machine-learning program, they can then build a system that monitors for the warning signs.
Ultimately, this could help scientists to stabilize plasma within the tokamak.
‘One thing that really excites me about the analysis technique I propose is that it is actually quite simple and could fairly easily be implemented by anyone who is developing these machine learning models,’ Parsons says.
‘All you have to do is take the numerical output of the prediction model, which in some sense describes how close you are to a disruption, change your inputs by a small increment, and compare the new output to the original output.
‘The smaller the change, the more stable the plasma discharge is with respect to the input variables. That is really the core of what I propose.’
The team from the US Department of Energy’s Princeton Plasma Physics Laboratory (PPPL) found that machine learning can be used to spot events that may bring on off-normal plasma behaviour inside the tokamak. ITER – the world’s largest fusion experiment – is shown above
The tokamak is the most developed magnetic confinement system and is the basis for the design of fusion reactors.
Plasma is contained in a vacuum vessel, which is then heated by driving a current through it.
A combination of two sets of magnetic coils creates a field in both vertical and horizontal directions, acting as a magnetic ‘cage’ to hold and shape the plasma.
The heating provided by the plasma current supplies a third of the 100 million°C temperature required to make fusion occur.
The tokamak is the most developed magnetic confinement system and is the basis for the design of fusion reactors. An illustration of the massive ITER device is shown
Additional plasma heating is provided when neutral hydrogen atoms are injected at high speed into the plasma, ionized and trapped by the magnetic field.
As they are slowed down, they transfer their energy to the plasma and heat it.
High-frequency currents are also induced in the plasma by external coils.
The frequencies are chosen to match regions where the energy absorption is very high.
In this way, large amounts of power may be transferred to the plasma.
Black-box models are typically shunned by the physics community, Parsons says.
But, the researchers says this type of approach could help to bring about solutions to these persistent problems.
‘As physicists, the way that we look at problems is trying to understand the relationship between what goes into your model and what comes out,’ he says.
‘It’s natural, then, that when we see these black-box models, we think that’s not something we want to deal with because we don’t understand what’s happening.
But, ‘a lot of the problems we’re facing in fusion are very technical, and if we could arrive at some of the solutions using machine learning, I think it’s prudent to explore all of the options and not exclude some just because they’re different from our training.’
Take a tour of the world’s largest nuclear fusion experiment