Simulation of an NSGA-III Based Fireball Inner-Temperature-Field Reconstructive Method

For the inner-temperature-field reconstruction of a fireball, a detecting method was proposed, using multi-channel visible spectral remote sensing theory. In our proposed method, the reconstructive algorithm based on multi-channel-detection was considered as a multi-objective optimization problem (MOOP), and a fast non-dominated sorting genetic algorithm based on reference-point strategy (NSGA-III) was employed as the solution of this problem. Besides, a so-called ambient pressure operator, based on the unique detecting model, was proposed and employed during the iteration process, for dynamic genetic parameter adjustment. To verify some performance of our proposed method, several numerical reconstructive simulations were carried out, from simple GA to NSGA-III, using several artificial 2-D virtual data, with different kinds of crossover and mutation functions. The simulation results show that, limited to our problems, the NSGA-III can effectively reconstruct different 2-D data, by a fixed crossover rate and a dynamic mutation rate under the proposed ambient pressure operator and an adaptive mutation rate function. The algorithm, limited to the field-reconstruction, also performs well on stability, but still has some deficiencies to be optimized.