A hybrid evolution optimization method based on a generative adversarial network model

Abstract

The invention discloses a hybrid evolutionary optimization method based on a generative adversarial network model, which mainly solves the problem that the traditional evolutionary algorithm is difficult to process high-dimensional and non-convex optimization and the like when facing an optimization problem, and comprises the following implementation steps of: (1) initializing a population; (2) calculating fitness values of the individuals according to a fitness criterion; (3) selecting dominant individuals; (4) performing crossover and mutation operation on the dominant individuals to obtainnew individuals; (5) taking the dominant individuals as samples, and generating new individuals by training the generative adversarial network model; (6) combining the new individuals obtained after the crossover and mutation operation with the new individuals generated through the generative adversarial network to form a new filial generation population; and (7) judging whether to terminate: outputting the optimal value of the target function after the algorithm is terminated, otherwise, returning to the step (2). According to the method, the global search capability and the convergence speedof the evolutionary algorithm are improved, and the method can be used for solving the complex high-dimensional optimization problem.

Description

technical field [0001] The invention belongs to the field of optimization design, in particular to a hybrid evolution optimization method based on a generative confrontation network model, which can be used to deal with complex high-dimensional optimization problems. Background technique [0002] As a general term for a class of optimization methods, evolutionary algorithms are inspired by the biological evolution mechanism of nature. Compared with traditional calculus-based methods and optimization algorithms such as exhaustive methods, evolutionary computing is a mature global optimization method with high robustness and wide applicability, and has the characteristics of self-organization, self-adaptation and self-learning , can effectively deal with complex problems that are difficult to solve by traditional optimization algorithms without being limited by the nature of the problem. Therefore, since the 1990s, research results related to evolutionary algorithms have spru...

AI Technical Summary

Problems solved by technology

However, the design of the probability model, the selection of the sampling method, and the precocity of the population evolution are also some problems that the distribution estimation algorithm needs to solve urgently.

Images