Machine learning characterisation of Alfvénic and sub-Alfvénic chirping and correlation with fast ion loss at NSTX

Abrupt large events in the Alfv\'{e}nic and sub-Alfv\'{e}nic frequency bands in tokamaks are typically correlated with increased fast ion loss. Here, machine learning is used to speed up the laborious process of characterizing the behaviour of magnetic perturbations from corresponding frequency spectrograms that are typically identified by humans. Analysis allows for comparison between different mode character (such as quiescent, fixed-frequency, chirping, avalanching) and plasma parameters obtained from the TRANSP code (such as $v_{\textrm{inj.}}/v_{\textrm{A}}$, $q$-profile, $\beta_{\textrm{inj.}}/\beta_{\textrm{A}}$). In agreement with previous work by Fredrickson \emph{et al.} [Nucl. Fusion 2014, 54 093007], we find correlation between $\beta_{\textrm{inj.}}$ and mode character. In addition, previously unknown correlations are found between moments of the spectrograms and mode character. Character transition from quiescent to non-quiescent behaviour for magnetic fluctuations in the 50 - 200 kHz frequency band is observed along the boundary $v_{\varphi} \lessapprox \frac{1}{4}(v_{\textrm{inj.}} - 3v_{\textrm{A}})$ where $v_{\textrm{inj.}}$ is the neutral beam injection velocity, $v_{\varphi}$ is the rotation velocity, and $v_{\textrm{A}}$ is the Alfv\'{e}n speed.