A neural network technique has been used to predict disruptions in the ADITYA tokamak. A time series prediction method is employed whereby a series of past values of some time dependent quantity is used to predict its value in the future. The time varying observables used in the present work are the different diagnostic signals from four Mirnov probes, one soft X ray monitor and one Hα monitor. The predicted quantities are the same observables at some future time. The neural network is trained with the past values of the different diagnostic signals as inputs and the future values of the same quantities as targets. The trained neural network is used to forecast in a multistep sequence. This amounts to a prediction several time steps earlier. Very good prediction results have been obtained up to 8 ms earlier with little distortion of the signals and no appreciable time lag, a capability which is believed to be well suited to the task of on-line predictions of disruptions in ADITYA. As actual experimental signals are used, confidence regarding the performance of the neural network on hardware implementation is automatically ensured.

An attempt is made to make a prediction of the disruption boundaries for the density limit disruption case using a neural network. Using experimental signals as input, the network should, in the long run, be able to provide information to the real time control systems about the density limit at which a discharge is likely to disrupt, so that the density can be kept below that limit. Several diagnostic signals are used from the ADITYA tokamak and are presented at selected time instants to the neural network inputs, in order to predict, at each of these instants, the density boundary. A disruption threshold has been established in order to examine the possibility of using the network as a real time disruption alarm. For most of the discharges this threshold is reached much before the actual disruption. The neural network is also used to make an optimization of the particular set of diagnostics in order to obtain the ones most crucial for predicting the density limit. The results of optimization have some of the features of the scaling laws of Murakami and Hugill. The optimized network compares well with the original one.

No abstract

No abstract

No abstract