Research And Application On Differential Evolution Algorithm

Differential Evolution(DE)algorithm is arguably one of the most powerful and versatile evolutionary optimizers for the continuous parameter spaces in recent times,and it is widely applied to many practical problems in different disciplines such as image processing,neural network and data mining due to its simple structure,easy implementation and rapid convergence.The DE algorithm has unique advantages,but the population evolves to the later stage,and the population diversity distribution decreases,leading to premature convergence of the algorithm.To address this critical issue,based on the standard DE algorithm,this paper carries out researches on the adoption of multi-population mechanism,weighted variation strategy,population covariance learning,adaptive control parameters,the solution of practical problems,and so on.The main research contents are as follows:Aiming at the contradiction between the late diversity of the DE algorithm and the slow convergence speed,Integration differential evolution algorithm with dynamic multiple population base on weighted strategies(MPWDE)is proposed.n the initial stage of population evolution,the whole evolutionary population is divided into multi-subpopulations,and different mutation strategies are adopted to improve the population diversity of the evolution process.The dynamic incremental parabolic crossover probability factor is used to increase the population diversity of the evolution process.The strategy weighting mechanism is adopted to improve the double mutation strategy to the weighted single mutation strategy to improve the convergence speed of the algorithm.The numerical simulation results show that the comparison effect between MPWDE and other classical improved DE algorithms is more significant.The diversity of the population and the crossover operator algorithm play an important role in solving global optimization problems in DE,the multi-poplutions covariance learning differential evolution algorithm(MCDE)is proposed.The population structure is a multi-poplutions mechanism,and each subpopulation combines the corresponding mutation strategy to ensure the individual diversity in the evolutionary process.The covariance learning establishes a proper rotation coordinate system for the crossover operation in the population.The adaptive control parameters balance the ability of population survey and convergence.Tested on the 25 benchmark functions of CEC2005,the experimental results show that MCDE has better optimization effect than other classical algorithms in solving global optimization problems.In the diagnosis and treatment of breast tumors,the accuracy of gland segmentation is directly related to the patient's treatment process,DE fuzzy entropy based on mammary gland segmentation is proposed.Combined with the image fuzzy entropy criterion,the evaluation function of the segmented gland is constructed.The image fuzzy entropy parameter is regard as the initial population of individual,after the mutation,crossover and selection of three evolutionary processes to search for the maximum fuzzy entropy of parameters,the optimal threshold of the segmented gland is achieved.The mammary gland is segmented by the threshold method of maximum fuzzy entropy.By comparing with other swarm intelligence fuzzy entropy algorithms in breast images,the experimental results show that this algorithm has higher accuracy in the segmentation of mammary glands.