Tokamak disruption alarm based on a neural network model of the high- beta limit

Abstract

An artificial neural network, combining signals from a large number of plasma diagnostics, was used to estimate the high- beta disruption boundary in the DIII-D tokamak. It is shown that inclusion of many diagnostic measurements results in a much more accurate prediction of the disruption boundary than that provided by the traditional Troyon limit. A trained neural network constitutes a non-linear, non-parametric model of the disruption boundary. Through the analysis of the input-output sensitivities, the relative statistical significance of various diagnostic measurements (plasma parameters) for the determination of the disruption boundary is directly assessed and the number of diagnostics used by the neural network model is reduced to the necessary minimum. The neural network is trained to map the disruption boundary throughout most of the discharge. As a result, it can predict the high- beta disruption boundary on a time-scale of the order of 100 ms (much longer than the precursor growth time), which makes this approach ideally suitable for real time application in a disruption avoidance scheme. Owing to the relative simplicity of the required computations, the neural network is easily implemented in a real time system. A prototype of the neural network disruption alarm was installed within the DIII-D digital plasma control system, and its real time operation, with a typical time resolution of 10 ms, was demonstrated

Open and emerging control problems in tokamak plasma control

“…Methods for predicting disruption onset have been developed (e.g. [34][35][36][37]), but almost all are based on training algorithms with data. Generally, a significant quantity of disruptive data is required for such training, which is likely to be difficult in ITER, which can tolerate only a small number of major disruptions before in-vessel different control objectives may be heavily coupled.…”

Application of Neural Networks for the Measurement of Electronic Temperature in Nuclear Fusion Experiments

“…MLPs have been successfully used in several nuclear research applications, such as plasma control [6][7][8], plasma parameter extraction [9][10][11], Gaussian fitting [12] and online detection of disruption precursors [13,14]. The MLP architecture with BP error learning is also used in the application described in this paper.…”

Detection of Disruptions in the High-β Spherical Torus NSTX

“…Neural networks are typically trained, in the sense that a predetermined sample of input and output data are used to determine the optimal values of the coefficients of the network. Neural networks have been used to predict various forms of disruptions on ASDEX Upgrade [84][85][86], DIII-D [87], ADITYA [88,89], TEXT [90,91], JET [85,92,93], and JT-60 [94,95].…”

“…Typical input data to the network include a measure of the plasma β (the normalized β (β N ) [96,97] or poloidal β (β P )), edge safety factor, plasma density (or Greenwald fraction [98,99]), locked mode amplitude, input power, radiated power, shape parameters, internal inductance, confinement time, and neutron emission. Some early work also used soft X-ray emission [87][88][89]91], high(er) frequency magnetic probes [87][88][89][90], or Dα monitors [87][88][89] though these diagnostics have typically not been used in more recent studies [85,86,92,93,95].…”