20 December 2017 Multiobjective generalized extremal optimization algorithm for simulation of daylight illuminants
Author Affiliations +
Abstract
Daylight illuminants are widely used as references for color quality testing and optical vision testing applications. Presently used daylight simulators make use of fluorescent bulbs that are not tunable and occupy more space inside the quality testing chambers. By designing a spectrally tunable LED light source with an optimal number of LEDs, cost, space, and energy can be saved. This paper describes an application of the generalized extremal optimization (GEO) algorithm for selection of the appropriate quantity and quality of LEDs that compose the light source. The multiobjective approach of this algorithm tries to get the best spectral simulation with minimum fitness error toward the target spectrum, correlated color temperature (CCT) the same as the target spectrum, high color rendering index (CRI), and luminous flux as required for testing applications. GEO is a global search algorithm based on phenomena of natural evolution and is especially designed to be used in complex optimization problems. Several simulations have been conducted to validate the performance of the algorithm. The methodology applied to model the LEDs, together with the theoretical basis for CCT and CRI calculation, is presented in this paper. A comparative result analysis of M-GEO evolutionary algorithm with the Levenberg–Marquardt conventional deterministic algorithm is also presented.
© 2017 Society of Photo-Optical Instrumentation Engineers (SPIE)
Srividya Ravindra Kumar, Ciji Pearl Kurian, Marcos Eduardo Gomes-Borges, "Multiobjective generalized extremal optimization algorithm for simulation of daylight illuminants," Journal of Photonics for Energy 7(4), 048001 (20 December 2017). https://doi.org/10.1117/1.JPE.7.048001 . Submission: Received: 30 July 2017; Accepted: 16 November 2017
Received: 30 July 2017; Accepted: 16 November 2017; Published: 20 December 2017
JOURNAL ARTICLE
16 PAGES


SHARE
Back to Top