Application Research On Multi-Strategy Multi-Swarm Slime Mould Algorithm

Slime Mould Algorithm(SMA)is a metaheuristic optimization algorithm that simulates the oscillation pattern of slime mould foraging in nature,which mainly simulates the behavior and morphological changes of slime mould during the foraging process.SMA has the advantages of fast convergence,strong expansibility,simple structure and easy implementation.Since its introduction,SMA has been widely used in various fields.However,SMA still has some shortcomings in solving large-scale and multi-objective optimization problems.Therefore,this thesis proposes some improved SMAs based on multi-strategy and multi-swarm and applies them to complex objective functions and engineering design,inverse kinematics and truss optimization problems to improve the search performance of SMA and broaden its application scope.The main work of this thesis is as follows:(1)In order to improve the convergence accuracy of SMA,balance the exploration and exploitation capability.Firstly,an Equilibrium Optimizer Slime Mould Algorithm(EOSMA)is proposed by combining the equilibrium optimizer operator and the random difference mutation operator with SMA.The equilibrium optimizer operator can help the algorithm achieve a better balance between exploration and exploitation,and the difference mutation operator can improve the probability of the algorithm jumping out of the local optimum in the late iteration.EOSMA was compared with other optimization algorithms on CEC2019 test suite and nine engineering design problems.Experimental results show that EOSMA has a strong competitive edge.(2)Aiming at the Inverse Kinematics(IK)problem of redundant manipulator,IK＿EOSMA and MOEOSMA are proposed based on the algorithmic framework of EOSMA for the IK problem in single-objective and multi-objective scenarios,respectively.Among them,IK＿EOSMA improves the difference mutation operator of EOSMA to improve the performance of the algorithm for the IK problem.MOEOSMA embeds Pareto archive and crowding distance operators on the algorithmic framework of IK＿EOSMA.The proposed algorithm was applied to the IK problem of a 7-DOF manipulator.Experimental results show that IK＿EOSMA and MOEOSMA can solve IK problems better.(3)Aiming at the large-scale multi-objective truss optimization problem,an Indicator-based Multi-swarm Slime Mould Algorithm(IBMSMA)is proposed.Pareto archiving strategy,shift-based density estimation indicator,chaotic grouping mechanism and dynamic regrouping strategy are introduced in IBMSMA.The shift-based density estimation indicator provides the selection pressure for population evolution.The chaotic grouping mechanism and dynamic regrouping strategy maintain the information exchange between populations,so that the algorithm to achieve a better balance between exploration and exploitation.IBMSMA was compared with the state-of-the-art algorithm on eight multi-objective truss optimization problems.Experimental results show that IBMSMA performs better in solving large-scale multi-objective truss optimization problems.