Research On The Improvements Of Collective Decision Optimization Algorithm

With the advancement of science and technology and the development of production technology,optimization problems almost cover all fields of scientific research and engineering practice,which makes optimization algorithm become an indispensable theoretical basis and research method of modern science and technology.Collective decision optimization algorithm(CDOA)studied in this paper is an evolutionary algorithm that simulates human collective decision behavior.It has the advantages of strong convergence,fast convergence speed and wide application range.However,CDOA performs well in depth search and poorly in breadth search,so it shows strong development capability and lacks good exploration capability.Therefore,aiming at the disadvantage that CDOA converges too fast and easily falls into local optimum,two improved CDOA algorithms are proposed to increase the optimization ability of the algorithm in solving global optimization problems,namely PCDOA and HCDOA.1)First,this paper combines an adaptive probability model with basic CDOA,and proposes a collective decision-making optimization algorithm based on adaptive probability(PCDOA).Specifically,in the novel search strategy,probabilistic P adaptively selects one of the two different mutation operators to perform mutation.The two mutation operators are one based on communication stage to enhance the diversity of population,and the other based on leader stage to accelerate the convergence rate.Through the interaction of two mutation operators,the development and exploration capability of the equilibrium algorithm are balanced.2)Second,another improved CDOA variant is the hybrid BSA collective decision optimization algorithm.A collective decision optimization algorithm based on historical information is proposed(HCDOA).In this algorithm,an adaptive reduced variable was designed.At that time,CDOA operator with strong development ability was selected to mutate.On the contrary,BSA operators with strong exploration ability are selected to mutate.Through the combination of the two algorithms,the mining and exploration capability of the balanced algorithm is achieved.Finally,two improved CDOAs are simulated on 23 benchmark functions and beams engineering design problems.The experimental results compared with other similar algorithms verify the effectiveness and competitiveness of the two improved CDOAs in practical problems.Comparing the two improved algorithms,the experimental data show that CDOA with BSA hybridization is more competitive than CDOA with self-improvement.