| The Grey Wolf Optimizer(GWO)is a group intelligence optimization algorithm inspired by the social hierarchy and hunting behaviour of wolves.Although GWO has received a lot of attention due to its simple structure and no extra parameters,it still has many shortcomings.The objective of this paper is to further improve the optimization mechanism of GWO and to solve the optimization problems in practical application scenarios,following the basic logic of scientific research of problem identification,problem analysis,problem formulation,problem solving and sustainable research.The main contents and results of this study are as below:(1)Firstly,in the process of researching GWO,a special phenomenon is found: when GWO solves the same optimization problem,its finding accuracy decreases as the distance between the theoretical optimal solution and the spatial location of the zero point increases.This paper defines the phenomenon as a specificity of GWO and develops an analysis of the reasons for this phenomenon.Furthermore,this paper proposes a set of shift test functions that can be freely adjusted to the location of the theoretical optimal solution.Through the analysis of the experimental results,the question is further raised: how to improve the performance of GWO for solving function problems in which the theoretical optimal solution is not located at zero point?(2)In response to these problems,research has been carried out to improve GWO.To determine an algorithm with optimal overall performance,each of the three perspectives was used to construct an improved grey wolf optimizer: No-Leader Cross Search Grey Wolf Optimizer(NLGWO),Stagnation Detection Based Grey Wolf Optimizer(SDGWO),Chaotic Oppositional Learning Based Grey Wolf Optimizer(COGWO).Testing the optimization accuracy of the above three algorithms through shift test function and CEC2019 test function.Testing the efficiency of the algorithms through CPU running time and time complexity analysis.The effectiveness of the improved algorithm was further verified by the Wilcoxon rank sum test.The results show that each of the three improved grey wolf optimizer has its own advantages,and all of them can improve the optimization performance of GWO to varying degrees for function problems whose theoretical optimal solutions are not located at zero point.Whereas SDGWO has the highest optimization accuracy,COGWO has the highest computing efficiency,and NLGWO has both higher optimization accuracy and computing efficiency,and has proven to be a simpler and more efficient algorithm with greater universality and is more suitable for replacing GWO to solve practical problems in different fields.(3)To solve optimization problems in practical application scenarios and to verify the effectiveness of NLGWO in practical problems,an application study of NLGWO is carried out.NLGWO is applied to the field of wireless sensing to address the problem of uneven distribution of nodes in the deployment of wireless sensor networks using the random deployment method.NLGWO is applied to the field of engineering design to address the difficulty and high cost of finding new concrete mixes using experimental methods.NLGWO is applied to the field of image processing to address the time consuming problem of the traversal method in the process of Kapur entropy multi-threshold image segmentation. |
PDF Full Text Request