Artificial intelligence system: will make unrestricted nuclear fusion reaction a reality

In recent years, researchers have been studying fission reactions that cause interruptions and damage to the tokamak installation. Currently, an artificial intelligence system capable of predicting and controlling fission reactions has been selected as the first project of the 'Aurora' supercomputer. One, the Aurora supercomputer is expected to arrive at the Argonne National Laboratory in 2021 and become the first mega-megacomputer system in the United States. Currently, artificial intelligence is working hard to study how to make the earth apply unlimited energy supply. It will eventually unravel the core. The mystery of fusion energy allows researchers to capture and control the processes that drive the sun and stars.

Researchers at the US Department of Energy's Princeton Plasma Physics Laboratory (PPPL) and Princeton University hope to use a huge new supercomputer to study how to use this donut-shaped device, the 'tokamaks'.

The computer system can achieve terabytes per second, 50-100 times faster than today's most powerful supercomputers. William Tang, chief research physicist at the Princeton Plasma Physics Laboratory, says: 'Our Research will use the deep learning of artificial intelligence to accelerate progress.'

This groundbreaking project will attempt to develop an experimentally validated method for predicting and controlling combustion plasma fusion systems such as ITER, which will validate the usefulness of fusion energy. It is reported that ITER is called 'International Thermonuclear Fusion'. The experimental reactor', also known as the 'artificial sun', is located in the southern small town of Cadarache in France. It is composed of 7 countries including the European Union, the United States, China, Japan, South Korea, India and Russia. TIER is also known as the most complex scientific project in human history.

Nuclear engineers at the ITER facility have now recruited a team of rocket scientists to help them produce super-strong materials that can withstand hotter temperatures than the sun. The ITER device is 5 meters in diameter and has a solid cross-section of 30 x 30 cm. ITER compression ring The huge magnet will be fixed in place.

The hydrogen plasma will be heated to 150 million degrees Celsius, 10 times higher than the sun's core temperature, allowing the fusion reaction to proceed. The fusion reaction takes place in a donut-shaped reactor called 'tokamak', which is made up of huge magnets. Surrounded by these magnets, they limit and circulate the superheated ionization plasma, keeping them away from the metal wall.

This superconducting magnet must be cooled to minus 269 degrees Celsius, as cold as interstellar space. For a long time, scientists have been trying to simulate the nuclear fusion process occurring inside the sun, which is believed to provide an almost unlimited amount of cheap, safe and clean power resources.

Unlike existing fission reactors, fission reactors will split strontium and uranium atoms, there is no risk of uncontrolled fusion chain reaction, and there is no long-term radioactive waste.

The deep learning software system developed by the Princeton Plasma Physics Laboratory is also known as the 'Recursive Neural Network Fusion System (FRNN)'. It is composed of a neural network, and users can train computers to detect events of interest through neural networks.

At the same time, this artificial intelligence 'recursive neural network fusion system' can quickly predict how large-scale tokamak plasma decomposes during fission reaction, and take effective control measures in time.

The overall goal of this study is to meet the challenging needs of the International Thermonuclear Experimental Reactor (ITER), which requires a prediction accuracy of 95% and a false alarm rate of less than 5%, at least 30 milliseconds or more before fission occurs. Time has happened.