Improvement Of Salp Swarm Algorithm And Its Application

Salp swarm algorithm(SSA)was proposed in 2017 by Mirjalili,an Australian scholar,to solve the unconstrained optimization problems.The algorithm SSA is a kind of a meta-heuristic swarm intelligent optimization algorithm.The salp swarm algorithm was inspired by the special predation method of salps that is like a chain of salps.Salp swarm algorithm has some advantages,such as simple structure,few parameters involved,and is widely used in many fields.However,in later iteration period,individual are prone to aggregation and falling easily into local optimum.Therefore,salp swarm algorithm still needs further research and improvement.The thesis mainly includes three aspects:A double-chain fractional order differential salp swarm algorithm DFSSA has been proposed to overcome the poor convergence and easy to fall into local optimum of the salp swarm algorithm.Algorithm DFSSA firstly introduces the double-chain strategy,that is,a leader is added.Two leaders are hired to make two chains of individuals perform different tasks,respectively.The first leader is always close to the food,and the second leader is close to the food in the early stage and away from the food in the later stage to avoid gathering.The fractional differential learning strategy is employed to increase the memorability of the algorithm DFSSA and make the algorithm jump out of the local optimum.The numerical experiments on 23 common standard test problems are carried out.The results show that the algorithm DFSSA has high solving accuracy,the required number of iterations to reach a given precision objective function value,and the ability to solve high-dimensional problems is superior to the comparison algorithm selected in this paper.The improved algorithm DFSSA is applied to PID controller parameter tuning,and a second-order system with delay is selected as the controlled object for simulation experiments.The simulation results show that the algorithm DFSSA has better ability to PID controller parameter tuning than the algorithm SSA.A double-leader salp swarm algorithm DLSSA coupling with loser elimination strategy has been proposed to overcome the shortcomings of salp swarm algorithm,which is easy to fall into local optimum and unable to obtain the objective function value meeting the accuracy requirements.The algorithm DLSSA employs two leaders whose tasks are similar to those of the two leaders in the algorithm DFSSA,and the followers in the population will randomly choose different leaders to follow.The loser elimination strategy is introduced to replace the weaker individuals in the population by new individuals,so as to maintain good population diversity.The numerical experimental results of 23 standard unconstrained optimization problems show that,compared to the salp swarm algorithm and other comparison algorithms,the algorithm DLSSA has better solving accuracy,requires fewer iterations to reach a given precision objective function value,and the ability to solve high-dimensional problems is stronger.The results of Wilcoxon test show that the algorithm DLSSA has significant advantages over the algorithm SSA.The algorithm DLSSA is applied to PID controller parameter tuning,and a second-order system with delay is selected as the controlled object for simulation experiment.The simulation experiment results show that the algorithm DLSSA has better performance than the algorithm SSA does in PID controller parameter tuning.The algorithm PF-DLSSA has been obtained by coupling the external point penalty function method with the algorithm DLSSA to solve the constrained optimization problem.Six standard constrained optimization problems are selected for numerical experiments.The numerical experimental results show that,the algorithm PF-DLSSA has a stronger ability to solve constrained optimization problems than the algorithm PF-SSA does,which is the algorithm combined salp swram algorithm with the external penalty function method.