A hybrid optimization algorithm and its application in flight trajectory prediction

Abstract

In order to improve the optimization performance of differential evolution (DE), a hybrid optimization algorithm (abbreviated as DEHHO) based on DE and Harris hawks optimization (HHO) is proposed. Firstly, the local search operator “HHO/SB” of HHO is combined with and classic mutation operator “DE/RAND” of DE to form a mutation link. Under the influence of the historical evolution state, each individual chooses a more suitable mutation operator to improve the possibility of successful evolution. Secondly, under the control of the historical evolution state, the updating of control parameters at the individual level assists the hybrid mutation operator to balance the population diversity and convergence rate during the evolution process. The performance of DEHHO is verified by a set of universal test benchmarks. On this basis, back propagation neural network (the initial parameters of which are optimized by DEHHO) is used to predict the flight trajectory, which further verifies the performance of DEHHO. Both validation results show that DEHHO outperforms other competitors under the same conditions.