Improved Gray Wolf Algorithm Based On Nonlinear Weight And Golden Sine Operator And Its Application

The swarm intelligence based optimization algorithm solves the optimization problem by simulating the intelligent behavior of the biological swarm to establish the model.Unlike exact approaches,the SI based algorithm obtains an initial solution through randomness and continuously improves this set of initial solutions during the optimization process.Therefore,the SI based algorithm has better local optimal avoidance and global exploration capabilities.At the same time,the non-derivation mechanism of the SI based algorithm treats the optimization problem as a black box,which can be used in a wider range of optimization problems.Gray Wolf Optimizer(GWO)simulates the predation process of gray wolf populations and constructs a mathematical model to solve optimization problems.In order to further improve the optimization accuracy and global convergence of the gray wolf algorithm,an improved gray wolf algorithm based on nonlinear adaptive weight and golden sine operator(NGS-GWO)is proposed.NGS-GWO first introduces non-linear weight,which enables search agents to adaptively explore and utilize the search space based on iteration,balancing the local exploitation and global exploration stages.Secondly,the golden sine operator is embedded in the gray wolf algorithm.Due to the special relationship between the sine function and the unit circle,traversing the sine function is equivalent to scanning the unit circle.The search agent conducts efficient search with the route of sine function,which improves the convergence performance of the algorithm and the global exploration ability.At the same time,the golden ratio coefficient makes the search agent indent by a fixed step each time.The search agent concentrates on the areas with excellent results for detailed exploitation,which improves the optimization accuracy and local exploitation ability.In the single objective function optimization part,the gray wolf algorithm(GWO),particle swarm algorithm(PSO),black hole algorithm(BH),firework algorithm(FWA),and sine cosine algorithm(SCA)are selected for comparison experiments with NGS-GWO.In the simulation experiment,other versions of the improved algorithm are also selected for comparison experiments with NGS-GWO to improve the persuasiveness of the comparison experiments.In addition,the effectiveness of the two improvement strategies has been verified by the control variable method.Finally,the improved algorithm is applied to high-dimensional optimization problems.The simulation results include non-parametric tests of statistical data and convergence curves.The results show that the introduction of the two improvement strategies can produce synergy,which effectively improves the optimization performance of the gray wolf algorithm.The algorithm is more competitive in terms of global convergence and local optimal value avoidance.NGS-GWO is also applied to optimization problems covering different backgrounds to further verify and improve performance.In the application optimization part,NGS-GWO is used as a trainer to apply a supervised learning multilayer perceptron model to solve multi-classification and function approximation problems.In addition,the adaptive threshold denoising method based on NGS-GWO is applied to ultrasonic testing.The Otsu multi-threshold segmentation method based on NGS-GWO is applied to gray image segmentation.The simulation results of the optimized design in practical application show that the improved algorithm is more competitive than other algorithms.