Application Research Of Particle Swarm Optimization Algorithm

With the development of computer technology,cloud computing,artificial intelligence,data mining and other technologies,more and more bionic algorithms have emerged,such as genetic algorithm,particle swarm optimization algorithm,bacterial foraging algorithm,hybrid frog hop algorithm,artificial bee colony algorithm,firefly algorithm,cuckoo search algorithm,fruit fly optimization algorithm and brainstorm algorithm,etc.Bionic algorithm,with its unique group intelligence behavior,has been increasingly applied to various fields of laboratory and practical engineering to solve complex problems including resource allocation,task scheduling and resource optimization.Due to its simple form,easy implementation and few parameters to be adjusted,particle swarm optimization(PSO)algorithm has been widely used in many disciplines and engineering fields.Therefore,It is of great practical value to study the applications of PSO algorithm and its improved version.This paper mainly focuses on particle swarm optimization algorithm and its improved algorithms to carry out two aspects of research:(1)Applying improved particle swarm optimization algorithm to select the optimal relaxation factor of the SOR iteration method.At present,the extant algorithm of selecting(SOR)optimal relaxation factor deeply depends on the determined segmentation strategy,that is,choosing the value of a split point as the relaxation factor and calculating the corresponding SOR iteration number,and then,return the value which corresponding SOR iteration number is less than the default SOR iteration threshold.However,this strategy is hard to find the global optimal one.In order to solve the problem above,we have proposed a stochastic strategy for searching the optimal relaxation factor in this thesis.The basic particle swarm optimization(bPSO),simple particle swarm optimization(sPSO),extremum disturbed Particle Swarm Optimization(tPSO),extremum disturbed and simple Particle Swarm Optimization(tsPSO)were used.The effectiveness and global optimality of the four algorithms in solving the SOR optimal relaxation factor problem are proved by combining the above four algorithms with the experiments of solving 5 sets of different linear equations.In addition,horizontal comparison has been made for the above four algorithms.Through experiments,it is proved that bPSO algorithm can quickly and efficiently converge to the optimal solution,but easily falls into the local optimal solution.The search range of sPSO algorithm in solution space is greatly improved compared with that of bPSO algorithm.tPSO algorithm and tsPSO algorithm can effectively overcome the drawback of the algorithm falling into local extremum.The above algorithms can well solve the problem of selecting the optimal relaxation factor of SOR iterative method.(2)Hybrid particle swarm optimization with greedy optimization and simulated annealing algorithm are proposed to solve 0-1 knapsack problems.In order to solve 0-1 Knapsack Problems more stably and efficiently,we propose a BPSOSA-CGOO algorithm which is organically composed by BPSO algorithm,greedy optimization strategy as well as the simulated annealing algorithm.Simulation experiments of 9 groups of different dimensions show that the BPSOSA-CGOO algorithm can solve 0-1 knapsack problems efficiently with small population size and iteration times.Meanwhile,the performed experiments of the algorithm can also find a better solution in 20-dimensions test data.Moreover,in the independent and repeated experiments,the BPSOSA-CGOO algorithm can hit the optimal solution with high probability for both low-dimensional and high-dimensional knapsack problems.The stability and reliability is significantly improved when using the BPSOSA-CGOO algorithm to solve high-dimensional knapsack problems.Through the above experiments and research,we have acquired the basic particle swarm optimization algorithm,the simple particle swarm optimization,the extremum disturbed particle swarm optimization and the extremum disturbed and simple particle swarm optimization,discrete binary particle swarm optimization algorithm,greedy optimization strategy,simulated annealing mechanism and etc.Through the application of the above algorithms,particle swarm optimization(PSO)and its improved algorithms show better optimization ability.Finally,according to the experimental results,we developed an application system for the above two functions,1)Using the tsPSO algorithm to select the optimal relaxation factor of SOR iterative method.2)Using BPSOSA_CGOO algorithm to solve 0-1 knapsack problem.