Particle leapfrog algorithms for high-dimensional nonlinear optimization problems
In response to the high-dimensional nonlinear optimization problem,through the study of intelligent algorithms such as APSO and GA,we constructed a newly improved intelligent algorithm by incorporating the ideas of adjacent compar-ison and moving repulsion.The algorithm constructs virtual particles by comparing the similar values of each particle(solu-tion)in different dimensions and calculates the fitness.In order to cope with the"premature"phenomenon,we take into consideration the randomness of particle jumping and the exclusivity of other particles in establishing the particle leapfrog updating formula;For the optimization problem of CEC2017 test function in different dimensions,PW is used together with APSO and GA for solution,and the superiority of the new algorithm is analyzed.The study shows that:(1)The new algo-rithm demonstrates strong superiority in the solution of high-dimensional problems with 10 kinds of test functions,and the excellence rate reaches 90%in the 30-dimensional solution;(2)In the spatial circle fitting problem,the fitness accuracy obtained by PW is 6.54%higher than that of APSO,and 6.93%higher than that of GA.The optimal solution is closer to the design value,with a maximum deviation of only 0.0864,meeting actual needs.It provides a new method for solving high-dimensional nonlinear optimization problems in engineering practice.
intelligent algorithmparticle swarm optimizationparticle warpgenetic algorithmspace circle fitting
国家科技期刊平台
NSTL
万方数据
