Solving Single-objective And Multi-objective Optimization Problems Based On Evolutionary Algorithm

Evolutionary algorithm is a kind of bionic random optimization algorithm,which seeks the optimal solution by simulating the process of biological evolution.This kind of algorithm generally does not require the objective function to be differentiable,which is more widely used than the traditional optimization algorithm,and has become a powerful tool to solve complex optimization problems.The optimization problem is widely existed in engineering practice and daily life.In this paper,improved algorithms for solving the corresponding problems are proposed,respectively.For a single objective optimization problem,this paper studies firefly algorithm(FA),and proposes a firefly algorithm based on topology improvement and crossover strategy.The traditional firefly algorithm has a slow convergence speed and is difficult to solve for high dimensional functions.In view of the above problems,this paper first uses the von neumann neighborhood structure to simulate the interaction of fireflies,thus reducing the computational complexity,improving the operation efficiency,and strengthening the global optimization ability of the algorithm,which is not easy to fall into the local optimization.Then,an adaptive crossover strategy is proposed,in which the optimal firefly of each generation can learn from other fireflies in different dimensions,so that the algorithm can jump out of local optimal effectively and find the global optimal solution more easily.Experiments show that the improved algorithm can improve the performance of the original algorithm significantly on most problems.In this paper,the inverse model-based multi-objective optimization algorithm(IM-MOEA)is studied,and AN-IMMOEA is proposed.In IM-MOEA,decision variables are randomly grouped to improve modeling efficiency.This method does not take into account the dependence between function and variables.In the improved algorithm,the neural network is used to evaluate the dependence between function and variables,so as to group variables and improve the reliability of the inverse model.In order to better deal with the problem of multiobjective optimization with irregular shape,an adaptive reference vector adjustment strategy is proposed.The reference vector is dynamically adjusted according to the distribution of each generation of non-dominant solutions and the search area is optimized.Experimental results show that compared with the original IM-MOEA algorithm and the existing multi-objective optimization algorithm with better performance,AN-IMMOEA proposed in this paper has better algorithm performance.