Study Of Distributed Brain Storm Optimization Algorithm And Its Application

In academia and industrial field,there are many complex optimization problems with the characteristics of non-continuous and non-differentiable.Meanwhile,the number of decision variables for these complex optimization problems is continuously increasing.For instance,in logistics services,there is a complex optimization problem—inventory routing problem(IRP).IRP involves the "trade-off" relationship between inventory and transportation,which is a type of NP hard problem.For these optimization problems,traditional methods are often unable to solve them effectively,but swarm intelligence optimization algorithms can be alternatives.The Brain Storm Optimization(BSO)algorithm is the "new star" of swarm intelligence optimization algorithms.It is an optimization algorithm based on human intelligence and inspired by the human brainstorming process in decision-making.It has a strong competitiveness among swarm intelligence optimization algorithms.Although BSO algorithm can get better results in solving optimization problems,but it has the same drawbacks as most swarm intelligent optimization algorithms,such as slow convergence and being easy to fall into local optimization.To fill the gap,BSO algorithm is investigated and applied to solving complex optimization problems,large-scale optimization problems and IRP.Firstly,the performance of BSO algorithm is investigated.BSO with a role-playing strategy(RPBSO)is proposed.In RPBSO algorithm,a role-playing grouping strategy is designed to improve the global search ability of the algorithm.An idea difference strategy is used to accelerate the speed of convergence of the algorithm and improve the local search capacity.The operation of re-initialization is introduced to get the algorithm out of the local optimum.The experimental results show the effectiveness of the RPBSO algorithm.Secondly,a novel distributed BSO algorithm is designed.In order to improve the time efficiency of the algorithm,a distributed BSO algorithm is proposed by using cloud computing technology.The distributed BSO algorithm is based on RDD model and uses a random grouping strategy.The feasibility of the distributed RGB SO algorithm is verified on the experiments of the large-scale optimization benchmark function of CEC 2010.And the speedup ratio of the distributed RGBSO algorithm is verified.Finally,BSO is applied to IRP.First,the mathematical model of the IRP is proposed.Then the RGBSO algorithm and the distributed RGBSO algorithm are used to solve the model.50 IRP examples are used to verify the two proposed BSO algorithms.The experimental results show that both algorithms are effective.It also follows that the performance of the distributed design scheme is the best.