Complementary search optimization method and system based on differential evolution

Abstract

The invention discloses a complementary search optimization method and system based on differential evolution, and the method comprises the steps: obtaining an initial population according to an initial parameter, carrying out evolution of at least one generation of the initial population, and obtaining a final solution. Based on the fact that the trend of the hyperbolic tangent function image isquite matched with the requirement degree of development in the evolution trend of the evolutionary algorithm, the use probability of the development search strategy is calculated through the extendedtanh, balance is achieved through the exploratory variation strategy with complementary characteristics, and therefore the optimization performance of the differential evolution algorithm is improved, and the performance of the differential evolution algorithm is improved. In addition, the complementary search optimization method and system based on tanh can well make up the blindness of search in the differential evolution algorithm and the premature phenomenon in the later period of evolution. The defect that in the prior art, proper strategies are difficult to select according to the characteristics of the current evolution stage is overcome. The invention is widely applied to the technical field of numerical optimization.

Description

technical field [0001] The invention relates to the technical field of numerical optimization, in particular to a differential evolution-based complementary search optimization method and system. Background technique [0002] The evolutionary algorithm can be used to maximize the desired attributes in the target and minimize the undesired attributes in the target, so as to realize the optimization of parameters, where the desired attributes and undesired attributes are determined according to the problem to be optimized. Among the existing evolutionary algorithms, there are usually many factors in the meta-heuristic algorithm that will affect the optimization results during the evolution process. For example, if an optimization process converges prematurely, it will easily fall into a local optimum and cannot jump out; Often leading to bumpy and spurious epistasis on the fitness surface can also mislead the search; neutrality and redundancy in the evolution process can slow ...

AI Technical Summary

Problems solved by technology

Among the existing evolutionary algorithms, there are usually many factors in the meta-heuristic algorithm that will affect the optimization results during the evolution process. For example, if an optimization process converges prematurely, it will easily fall into a local optimum and cannot jump out; Often leading to bumpy and spurious epistasis on the fitness surface can also mislead the search; neutrality and redundancy in the evolution process can slow down evolution, while also helping escape local optima since the entire search space is explored ; Due to its unique variation and crossover method, the differential evolution algorithm treats every member of the population equally during the evolution process without considering their fitness information and position information, which leads to a certain blindness in the early search. Gradual reduction of differential information among members also leads to premature convergence

Images