Electron density profiles of fusion plasmas may exhibit large variations due to different operating regimes as well as fast changes due to transient plasma phenomena (e.g. ELMs), which modify the frequency content of broadband reflectometry signals. For this reason, the evaluation of the density profile based on the spectrogram technique usually requires: (i) the use of conservative sizes of the window of analysis, in order to cope with the large/fast data variations, or (ii) direct user intervention to match the window size to the data content. In the first case, accuracy is reduced, while in the second, the need for user intervention prevents the fully automatic evaluation of the density profile. Here we present the application of a new technique that uses a time-frequency energy concentration measure in order to estimate the optimum window length. In this way, the spectrogram analysis may be automatically adapted to the frequency content of the data in each window, increasing the time-frequency resolution in cases where the frequency exhibits fast/large variations, as for example just inside the pedestal region of H-mode plasma profiles. The application to test signals and simulated profile data shows the mportance of the new technique in improving the accuracy of automatically evaluated density profiles. The advantages of the technique versus its computing requirements are also discussed. |
Trabalhos Científicos 
Comunicações a Reuniões Científicas