A Two-Stage Differential Evolution Algorithm with Mutation Strategy Combination
10
Citation
47
Reference
10
Related Paper
Citation Trend
Abstract:
For most of differential evolution (DE) algorithm variants, premature convergence is still challenging. The main reason is that the exploration and exploitation are highly coupled in the existing works. To address this problem, we present a novel DE variant that can symmetrically decouple exploration and exploitation during the optimization process in this paper. In the algorithm, the whole population is divided into two symmetrical subpopulations by ascending order of fitness during each iteration; moreover, we divide the algorithm into two symmetrical stages according to the number of evaluations (FEs). On one hand, we introduce a mutation strategy, DE/current/1, which rarely appears in the literature. It can keep sufficient population diversity and fully explore the solution space, but its convergence speed gradually slows as iteration continues. To give full play to its advantages and avoid its disadvantages, we propose a heterogeneous two-stage double-subpopulation (HTSDS) mechanism. Four mutation strategies (including DE/current/1 and its modified version) with distinct search behaviors are assigned to superior and inferior subpopulations in two stages, which helps simultaneously and independently managing exploration and exploitation in different components. On the other hand, an adaptive two-stage partition (ATSP) strategy is proposed, which can adjust the stage partition parameter according to the complexity of the problem. Hence, a two-stage differential evolution algorithm with mutation strategy combination (TS-MSCDE) is proposed. Numerical experiments were conducted using CEC2017, CEC2020 and four real-world optimization problems from CEC2011. The results show that when computing resources are sufficient, the algorithm is competitive, especially for complex multimodal problems.Keywords:
Differential Evolution
Premature convergence
Recently,the use of evolutionary algorithms(EAs) to solve the Multi-objective Optimization Problems(MOPs) has attracted much attention.EA is a population based optimized approach which can find a group of Pareto-optimal solutions in a single run.Differential Evolution(DE) is a branch of EA that is developed to handle problems over continuous domains.An improved Multi-objective Evolutionary Algorithm is proposed based on Differential Evolution(CDE) to solve MOPs.The proposed algorithm is compared to the other two classical Multi-objective Evolutionary algorithms(MOEAs) NSGA-Ⅱ and SPEA2 with the experiment results.
Differential Evolution
Cite
Citations (1)
Differential evolution, termed DE, is a novel and rapidly developed evolution computation in recent year. There are some advantages of DE, including simple structure, easy use and rapid convergence speed. Besides, DE can be also applied on complex optimization problem. However, there are some problems, such as premature convergence and stagnation, remaining in DE algorithm. To overcome those disadvantages, a different method was proposed, named CO-DE, by combining with a simple co-evolutionary model and reset mechanism. Thus, CO-DE can maintain appropriate swarm diversity and reduce the premature convergence. On the other hand, a reset mechanism was set to avoid the particle stagnates, which can further improve the performance of differential evolution. The proposed model can be now successfully applied with some well-known benchmark functions.
Premature convergence
Differential Evolution
Benchmark (surveying)
Reset (finance)
Cite
Citations (0)
Differential evolution (DE) is a fast and effective computing method and technique. In differential evolution for global optimization, mutation plays a key role in the performance and there are several mutation variants, which have been widely used in both benchmark test functions and real-world applications. However, most of these mutation variants can only generate one offspring in one mutation operation. In order to make the best of the information of multiple parents in the process of mutation, this paper proposes a multi-parent mutation, and then extends differential evolution with the multi-parent mutation to handle multi-objective optimization problems. Simulation results on a set of test functions show that the proposed approaches can improve the search performance.
Differential Evolution
Benchmark (surveying)
Global Optimization
Cite
Citations (1)
Reduction of population diversity leads to premature convergence,which limits search capability and computational efficiency of evolutionary algorithm.To deal with premature convergence,the evolutionary population is updated with elite solutions and new created random solutions periodically during evolutionary process.Adding elite solutions means inheriting results got by anterior evolutionary process from the beginning and adding new solutions created randomly improves population diversity.For large search scale,population is remodeled many times in the whole evolutionary process according to the search scale.To test solution quality and computational efficiency,the proposed remodeling population strategy is applied to symbiotic evolutionary algorithm for dealing with a flexible job-shop scheduling problem.Compared with the widely used traditional evolutionary algorithm,improved algorithm shows better performance for different search scale no matter whether the problem is large or not.The most important is it presents a solution for dealing with premature convergence,which deeply limits performance of evolutionary algorithm.With the remodeling population strategy,the applying depth and width of evolutionary algorithms will be improved.
Premature convergence
Cultural algorithm
Cite
Citations (0)
Compact Genetic Algorithm(CGA) requires a small amount of memory,but it is apt to premature stagnate.This paper proposes a Mutation-Based Compact Genetic Algorithm(MBCGA) by introducing the mutation operator into CGA,thus MBCGA mimics all the main genetic operators in natural evolution,then local search is strengthened and premature stagnation can be avoided.Experimental results show that the MBCGA generally exhibits a higher rate of convergence than CGA,without increasing the memory requirement.The effect of the introduced mutation operator is analyzed and verified.
Premature convergence
Operator (biology)
Genetic operator
Cite
Citations (0)
Differential Evolution
Evolutionary Dynamics
Constant (computer programming)
Cite
Citations (1)
Premature convergence is an important factor affecting optimization results in genetic algorithms(GA) . Effect on sample variety of mutation probability delivered by experience value in mutation method is analyzed, and self adaptation mutation method controlling mutation operator is advanced and how to select mutation time and mutation probability based on evolution process is dealt with. The comparison of the results from actual example calculations has proved that the improved self adaptation mutation algorithms can solve the problem of premature convergence effectively.
Premature convergence
Adaptive mutation
Operator (biology)
Cite
Citations (0)
The standard differential evolution is easy to fall into premature convergence; so for solving the problem, the paper presented a Multi Population Differential Evolution Algorithm (MPDE) which was based on simulating the developing history of human races. The population was divided into several subpopulations in MPDE, which can enhance the searching capacity of the algorithm. Finally, three experiments were done on benchmark functions, the results show that MPDE has a good global searching capacity than DE and could avoid premature convergence.
Premature convergence
Differential Evolution
Benchmark (surveying)
Cite
Citations (0)
Premature convergence
Differential Evolution
Cite
Citations (62)
Although BOA is effective at finding solutions for optimization problems, small population sizes in a model can result in premature convergence to a sub-optimal solution. One way of avoiding premature convergence is to increase population diversity with a mutation operator. In our experiments, we compare several mutation operators for use with BOA. We examine in detail the probabilistic model utilizing (PMU) bit flipping mutation operator. We compare the effectiveness of the PMU operator with standard BOA, self-adaptive evolution and local search of substructural neighborhoods.
Premature convergence
Operator (biology)
Adaptive mutation
Cite
Citations (1)