The new code, the Fusion Recurrent Neural Network (FRNN), applies deep learning to a pair of comprehensive databases at the world’s largest nuclear energy facilities.
These facilities contain historic logs of detailed data on the disruptions that occur in fusion energy development. The new FRNN has demonstrated the ability to mitigate these disruptions to virtually zero, providing limitless capabilities for nuclear energy creation.
After applying deep learning to the data sets, The FRNN can reliably forecast disruptions and destructive events that occur in tokamaks.
The tokamak, invented by the Soviet Union in the 1960s, is the most advanced nuclear energy generator. The tokamak is a type of magnetic confinement device that uses magnetic fields to trap hot plasma.
Two magnetic field coils, the toroidal and poloidal, trap the plasma in a vertical and horizontal field, shaping the hot plasma and keeping it off the machine walls. The plasma, trapped in the vacuum vessel, is brought to life with a puff of smoke and a strong electric current. From there, three sources of heat energy are applied to ultimately create enough heat to generate fusion. Neutral hydrogen atoms are injected into the plasma at high speed.
As they are slowed, they transfer energy to the plasma. Secondly, plasma current is induced using a transformer and a central magnetic coil. Lastly, radio-frequency heating is applied using external coils that deliver oscillating currents where the energy absorption is very high. With enough heat energy, fusion can occur, a process in which two atoms of hydrogen fuse into one atom of helium. The leftover mass from the fused hydrogen is converted into energy. This is akin to the process that powers the sun...Read The Full Article Here>>>...