Identification of the Sensitive Sensors in ADITYA Tokamk

Position and shape of plasma current channel is determined by magnetic measurements in tokamak, a fusion device. To avoid the direct contact of hot plasma with the surrounding vacuum vessel, the position of this plasma current channel is controlled in a millisecond time-scale by feedback control system. It is therefore important to ensure the reliability of the magnetic sensors used to determine position and shape parameters. The sensitivity of each probe to these parameters is estimated by designing a physical model based on a multi-layered feed forward artificial neural network. Input to this network is simulated magnetic signals and the outputs are the position and shape parameters. In this paper first the neural network is trained with simulated data and later the identification of the sensitive sensor is done.