Enhanced Grey Wolf Optimization Algorithms Based On Local Parameter And Population Diversity Improvement

Swarm intelligence optimization algorithms provide new ideas for optimization problems of non-convex and complex systems with non-linear,high dimensionality and multiple constraints by virtue of their population's ergodicity,randomness,and adaptability.As an emerging meta-heuristic algorithm,grey wolf optimization algorithm has the characteristics of simple structure and flexible search.However,there are problems such as falling into local optimality and low convergence accuracy in the optimization process of high-dimensional complex systems.Therefore,this thesis will further improve the grey wolf algorithm based on its advantages of fast convergence,flexible and adjustable parameters,and simple structure.The main research contents of this thesis are as follows:1.In order to strengthen the effect of local parameters in the standard grey wolf optimization algorithm,and improve universality,convergence accuracy and convergence speed of the grey wolf optimization algorithm,this thesis proposes two enhanced strategies:(1)Since a single convergence strategy cannot perform well in a variety of optimization problems,a piecewise nonlinear convergence strategy with a nonlinear adjustment factor is proposed,and it is verified that it has better universality and higher convergence accuracy.(2)Considering that the original position updating formula cannot highlight the leadership ability of different elite wolves,a new elite wolf position updating strategy based on the random opposition learning method is proposed,and this strategy is proven to be helpful to improve local search ability.2.In order to improve the population diversity of the wolf group in the whole iteration process of the standard grey wolf optimization algorithm,this thesis proposes two strategies to improve the convergence speed,global convergence ability and robustness:(1)As randomly distributed initial population cannot well represent the characteristics of the search space,an initialization method for population distribution based on the improved good point set theory is proposed,which improves the diversity of the initial population,and is verified that it does improve the algorithm's convergence speed.(2)Aiming at the issue of losing population diversity in the whole iterative process,a fusion strategy using the rule of survival of the fittest to dynamically evolve populations is proposed,and it is verified that it has a stronger global convergence ability,and at the same time it has a better performance in solving constrained optimization problems with strong competitiveness and high reliability.