System Identification for EAST plasma shape and position control

摘要

The system identification algorithm can build a device-specific control-oriented plasma response model by using measurement data. The EAST shape and displacement control system includes the shape control, slow vertical position (slow Z) control and fast vertical position (fast Z) control. The shape and slow Z are controlled by poloidal field (PF) coils while the fast Z is controlled by in-vessel coils (IC). Based on the algorithm of the EAST shape and slow Z control, we use the system identification method to build a control-oriented plasma shape model with PF coil currents as inputs and flux errors at control point and slow Z error as outputs. In order to obtain better response information, a large number of closed-loop experiments have been designed using 12 independent PF coils. For each experiment in a set, different PF coil-inputs were excited with a suitable input signal containing different frequencies and amplitudes. The experimental results have been analyzed by the system identification method with the Auto-regressive moving-average exogenous input (ARMAX) model and cross-checking has been performed between different shots. The identified models are compared with some physics based models.