Research On Multi-population Integration Constrained Differential Evolution Algorithm

Constrained optimization problem is a common problem in the fields of engineering technology,mathematics,operations research and computer science.Compared with traditional constrained optimization methods,evolutionary algorithm has the advantages of maintaining population diversity and avoiding falling into local optimal solution,it is widely used to solving constrained optimization problems.Differential evolution algorithm is a random population search algorithm for solving nonlinear,high-dimensional and some other complex optimization problems,which performs well in the first evolutionary algorithm competition.According to the no-free lunch theorem,it is difficult to obtain the global optimal solution at the same time by using a single differential evolution strategy or a single constraint processing technique to solve constrained optimization problems with different performance.In view of these shortcomings,this paper integrates several improved differential evolution algorithms and constraint processing techniques,and proposes an improved constrained differential evolution algorithm.The research contents of this paper are as follows:1.In order to overcome the deficiencies of premature convergence and low optimization accuracy,an adaptive constrained differential evolution algorithm based on probability interval update mechanism is proposed.First,a given continuous range is divided into several intervals and obtains the probability.Two continuous values as the mutation probability are generated according to the two initial values of the interval.Thus,the algorithm can balance the global search ability and the local search ability.Second,the new population is selected by the constraint processing technique,and the restart mechanism is used to jump out of the local optimal solution and to increase the diversity of the population.Then,an adaptive parameter control mechanism is introduced to enhance the robustness and adaptability of the algorithm.The numerical simulation results show that the improved constrained differential evolution algorithm not only maintains the advantages of each mutation operator,but also improves the convergence speed of the algorithm.2.In view of lack of global search ability when solving complex optimization problems,and making full use of the advantages of multi-population evolution model,a constrained differential evolution algorithm based on multi-population framework is proposed.The algorithm is integrated by three improved differential evolution algorithms.First,the whole population is divided into three smaller subpopulations with the same size and a larger reward subpopulation.Second,each of the three improved differential evolution algorithms has a subpopulation,and the reward subpopulation is assigned to the differential evolution algorithm with the best performance as an additional reward,which can achieve information sharing and close cooperation among variables,and the population size is redistributed according to the improved proportion of the fitness value every certain generations.Then,the population is updated by the constraint processing technique.The algorithm is tested on three kinds of constrained optimization problem sets.The experimental results show that the algorithm proposed in this section can effectively solve the constrained optimization problem.3.In view of the lack of constraint processing ability of the single constraint handling technique,a constrained differential evolution algorithm based on ensemble of constraint handling techniques and multi-population framework is proposed.First,through the constraint allocation mechanism,two constraint handling techniques and three difference improved algorithms into constraint algorithms are dynamically matched to form a pool of constraint algorithm combination.Second,the population is divided into three smaller subpopulations and one larger reward subpopulation.Then,a combination of constraint algorithms is randomly selected from the combined pool,and the three constraint algorithms are run in three subpopulations respectively.According to the improvement of the fitness value,the optimal constraint algorithm is selected to run on the reward subpopulation.The test function test shows that the algorithm can effectively improve the overall performance of integrated differential evolution algorithm in solving constrained optimization problems.