Current control of EAST fast control power supply based on single neuron adaptive predictive PI

The Experimental Advanced Superconducting Tokamak (EAST) is an important device developed by China to realize controllable nuclear fusion. The vertical displacement of plasma is controlled by the current output from EAST fast control power supply. Traditional PI control has shortcomings in current dynamic response. A single neuron adaptive predictive PI (SNAD-PPI) control algorithm is proposed to ensure fast and accurate current output. The adaptive structure is used combined with the single neuron structure to improve the problem of insufficient dynamic response under the control of fixed PI parameters. Advance tracking reference current is used as the adaptive objective function of tutor-supervised learning for setting PI parameters in a single neuron. The proportional adjustment coefficient and the proportional learning rate of single neuron adaptive control are adjusted in real time to realize fast convergence of single neuron. Outstanding current dynamic response performance compare to traditional PI control is verified by experiments.