Abstract:Collective Thomson scattering (CTS) is a diagnostic technique that obtains ion temperature and ion composition of plasma by spectral decomposition from scattering spectra. Bayesian estimation and least squares fitting are usually applied in this spectral decomposition process. Nevertheless, these spectral decomposition methods strongly rely on measurements of other diagnostic systems, and the measurement errors of other systems would influence the spectral decomposition results. In this article, an improved genetic algorithm is applied to decompose the scattering spectra of CTS. By analyzing the sensitivity of plasma parameters, the width and slope of the scattering spectrum are found to be strongly associated with ion temperature. Based on this correlation relation, a new fitness function is designed to provide a more precise estimation of ion temperature. Meanwhile, adaptive crossover and mutation operators are introduced to solve the premature convergence problem. This improved genetic algorithm with the new fitness function can obtain a more precise ion temperature from scattering spectra of CTS and does not rely on the measurement of other diagnostic systems, which has an extensive application prospect in data processing of CTS.

Spectral decomposition for collective Thomson scattering based on an improved genetic algorithm