Intelligent control for predicting and mitigating major disruptions in magnetic confinement fusion

Abstract

Magnetic confinement fusion is believed to be one of the promising paths that provides us with an infinite supply of an environment-friendly energy source, naturally contributing to a green economy and low-carbon development. Nevertheless, the major disruption of high temperature plasmas, a big threat to fusion devices, is still in the way of mankind accessing to fusion energy. Although a bunch of individual techniques have been proved to be feasible for the control, mitigation, and prediction of disruptions, complicated experimental environments make it hard to decide on specific control strategies. The traditional control approach, designing a series of independent controllers in a nested structure, cannot meet the needs of real-time complicated plasma control, which requires extended engineering expertise and complicated evaluation of system states referring to multiple plasma parameters. Fortunately, artificial intelligence (AI) offers potential solutions towards entirely resolving this troublesome issue. To simplify the control system, a radically novel idea for designing controllers via AI is brought forward in this work. Envisioned intelligent controllers should be developed to replace the traditional nested structure. The successful development of intelligent control is expected to effectively predict and mitigate major disruptions, which would definitely enhance fusion performance, and thus offers inspiring odds to improve the accessibility of sustainable fusion energy.

Keywords

• magnetic confinement fusion
• intelligent control
• major disruption