一种实现全父辈交叉遗传算法的神经计算模型

A Neural Computing Model for Entire Parens Recombination Genetic Algorithm

该文提出了一种实现二值编码全父辈交叉遗传算法的神经计算模型GAnn。GAnn将遗传算法的迭代循环结构通过一组由神经元模块所组成的反馈回路实现,并给出了实现二值编码全父辈交叉操作以及突变操作的人工神经元和神经网络拓扑结构。该文通过实验结果验证了GAnn的可行性。GAnn综合了硬件遗传算法和并行遗传算法的优点,这对于用硬件实现遗传算法,显式地实现遗传算法的内在并行性,提高遗传算法的实时性,拓宽遗传算法应用领域的研究具有积极的意义。

This paper proposes a neural computing model for the binary coded Entire Parents Recombination Genetic Algorithm for the sake of integrating the hardware genetic algorithm and the parallel genetic algorithm.In the model,the evolving iteration of GA is implemented with a feedback loop composed of neural network modules.The neural networks as well as the neurons of both binary coded entire parents recombination and mutation operation are designed.Simulations on a suit of benchmark functions validate the model.This research may lead a new way of improving the real-time performance of GA,which not only implement GA with hardware but also realize the inherent parallelism of GA explicitly.