A quench occurs when part of a superconducting magnet suddenly shifts from the superconducting state to the normal state. If it is not detected promptly and protective measures are not taken, the magnet can overheat and be damaged.
A team at the Institute of Modern Physics of the Chinese Academy of Sciences proposed an online detection method based on deep learning. It addresses false triggers and increased detection delays that can affect conventional methods amid strong magnetic flux jumps.
The researchers trained a convolutional neural network (CNN) using detection data accumulated over several years. The model identifies subtle differences between magnetic flux jumps and quench signals. They then developed and deployed an artificial intelligence quench detection system, AI-QDS, for real-time online monitoring.
