ONLINE CONTROL OF THE COMPASS-D TOKAMAK USING A NEURAL-NETWORK

Multi-layer perceptron (MLP) networks are particularly appropriate for performing rapid non-linear mapping, In the application discussed in this Paper the position and shape of the plasma within the experimenta l fusion research tokamak COMPASS-D at UKAEA's Culham Laboratory is de termined from a series of magnetic sensors placed around the vacuum ve ssel, close to the plasma boundary. By using a real-time analogue neur al network it is possible to achieve control within a sub-millisecond time-scale. In this application the neural network is needed to solve an inverse problem. Numerical codes exist that are able to calculate t he signals expected on the magnetic sensors for a given plasma positio n and profile. The problem is well defined from the solution of the Gr ad-Shafranov equation. However, no easy analytical formalism exists to reverse the problem - to calculate the plasma parameters given the ma gnetic signals. It is this mapping, from the set of magnetic diagnosti c input signals to the parameters of the plasma, that an MLP network c an be trained to solve. The training data are some 2000 example plasma equilibria, covering the likely possible configurations of the plasma , solved by numerical methods. The desired aim, to control the plasma boundary position to within a few millimetres, has now been achieved.