Generalized Extremal Optimization Algorithm to design a LED-based spectrally tunable light source for Star Simulation

An Autonomous Star Tracker (AST) has been designed and is being constructed at National Institute for Space Research (INPE). In order to calibrate and characterize the AST a sophisticated infrastructure of tests is required. This infrastructure, composed of several instruments, has included one Star Simulator (SS). The simulator has also been designed and constructed at INPE. The SS is basically composed of a LED-based spectrally tunable light source, a pinhole and a collimator. The SS is designed to have the capability of producing different continuum spectral distributions, mimicking star magnitudes in the visible and near-infrared range, what is achieved through the feedback control of individual LEDs. The objective of this work is to present a methodology for the optimal selection of the LEDs that compose the light source. For that, an evolutionary multi-objective approach is used. This approach tries to get the best spectral simulation and at the same time to reduce the number of LEDs used. The methodology applied to model the stars is also presented in this paper, together with a theoretical basis applied to model the LEDs. A series of simulations have been conducted to predict the performance of the designed tunable source. Many source distributions have been constructed for a number of target distributions, and the results are promising.