Researches On Improvement And Application Of Seagull Optimization Algorithm

Seagull optimization algorithm(SOA)is an intelligent optimization algorithm based on population,which is inspired by the migration and hunting behavior of seagulls.It is characterized by simple structure,strong adaptability,and fast convergence speed.And there are some defects of the algorithm,such as evolutionary mode,dependence on the optimal individual,parameter control and so on.Based on the previous work,the corresponding improvement strategy is proposed,and the improved algorithm is applied to UAV trajectory planning and BP neural network optimization.The main contents of this thesis are as follows:1)An Adaptive Mutation Seagull Optimization Algorithm(ASOA)is proposed from the perspective of search mode optimization.Aiming at the single search mode and over-dependence on the best individual of SOA,a population evolution strategy based on multiple impact factors is proposed,and an adaptive mutation strategy for the best individual is introduced to improve the populations diversity that make it easier for seagulls to escape local optimal traps.Also,a cosine function-based parameter controlling is adopt to balance global and local search of ASOA.The effectiveness of the algorithm is verified by the optimization experiment of standard test functions.2)A Double-group-based Seagull Optimization Algorithm(DSOA)is proposed from the perspective of algorithm structure optimization.Aiming at the poor local search ability and over-dependence on the best individual of SOA,a dual-group coevolution strategy based on task assignment is proposed for DSOA,and a dynamic feedback adjustment mechanism is introduced to adjust the proportion of the two subgroups.Besides,a population regeneration mechanism based on random walking is adopt to improve the populations diversity.The effectiveness of the algorithm is verified by the optimization experiment of standard test function.3)A Hybrid Seagull Optimization Algorithm(HSOA)is proposed.The hybrid improvement is based on DSOA,which integrates the improvement strategy of ASOA and sparrow search algorithm(SSA).The HSOA has further improved the optimization performance of standard test function,and it is applied to unmanned aerial vehicle(UAV)trajectory planning and BP neural network parameter optimization for the verification of practicability.