Study For Multiobjective Intelligent Optimization Algorithm

In engineering applications and scientific research,it is often necessary to optimize multiple objectives simultaneously.In view of these problems,the traditional mathematical programming method has low efficiency,while the intelligent optimization algorithm has achieved good results in solving the multi-objective optimization problem.With its flexibility and versatility,it has a great development prospect and research value.Therefore,this paper proposes three new multi-objective evolutionary algorithms,and the main contents of this paper include the following three parts:(1)In the field of evolutionary multi-criteria optimization,the hypervolume index is the only known strict monotone unitary measure in Pareto dominance.However,the amount of computation required to calculate the hypervolume hinders the potential of using this index.Therefore,a grid-based Counts multi-objective Evolutionary Algorithm(GC-MOEA)is proposed in this paper to replace the accurate hypervolume value by counting the grid points generated.In detail,what we often need is not the exact hypervolume value,but its strict monotonic characteristic in the aspect of optimality,so as to find another index with substitution.Therefore,grid counting is used to replace the hypervolume,and the hypervolume index is applied to the evolutionary algorithm as the fitness function of the population.(2)As a new metaheuristic algorithm proposed in recent years,crow search algorithm has shown great potential in multi-objective optimization.However,the existing multi-objective crow search algorithm can only solve the unconstrained optimization problem,but the actual engineering problems often have constraints.In order to expand the scope of its application,a new Constrained Multi-objective Crow Search Algorithm(C-MOCSA)was proposed.The algorithm adopts multi-objective method and n-dimensional spherical search to guide the search direction of points based on the original crow search algorithm.At the same time,a new dominant principle of infeasible solution is proposed,which can make full use of the constraint value information of all infeasible points to guide the search from infeasible solution to feasible region.In addition,the algorithm adopts global search and local search in parallel,which can not only strengthen the local search ability of individuals in the nearby region,but also can add new individuals to avoid the population falling into the local optimal solution.(3)Grid has the advantage of reflecting both convergence and diversity,but the existing references have not fully utilized the potential of the grid.In this paper,a new Double grid-based multi-objective Search Algorithm(DG-MOSA)is proposed,which divides the target space and the decision space into grids,and the grid-based method is used to enhance the searching ability of the population toward the optimal direction,while maintaining a scattered and uniform distribution among the optimal approximate solutions.In addition,in the process of environment selection,the algorithm proposes a dual dominance ranking method.Based on each dominance depth after rapid nondominance ranking,grid dominance is used again for each population depth to provide higher selection pressure for the population.