基于遗传算法优化权重的组合预测分析
A Study of Combination Forecasting Optimized by Genetic Algorithm Weights
黄胜龙 1袁宏俊 1胡凌云2
作者信息
- 1. 安徽财经大学统计与应用数学学院,安徽 蚌埠 233030
- 2. 安徽财经大学管理科学与工程学院,安徽 蚌埠 233030
- 折叠
摘要
目的提出了一种利用遗传算法优化组合预测权重的方法,旨在通过遗传算法的优化过程选择最佳权重,以提升组合预测模型的精度.方法首先,采用网格搜索法在超参数格中寻找适宜的遗传算法参数.随后,通过遗传算法的自然选择、交叉和变异等操作,逐步演化出适用于组合预测的最佳权重.最后,通过对全国铁路运营里程的实证分析.结果结果显示基于遗传算法确定的权重在组合预测中表现出更卓越的预测效果,相较传统方法具备更高的准确性和泛化能力.
Abstract
This paper proposes a method for optimizing combination forecasting weights using genetic algorithms,aiming to en-hance the accuracy of combination forecasting models by selecting optimal weights through the optimization process of genetic algo-rithms.Firstly,a grid search method is employed to explore suitable genetic algorithm parameters within the hyperparameter space.Subsequently,through operations such as natural selection,crossover,and mutation in genetic algorithms,optimal weights applica-ble to combination forecasting are gradually evolved.Finally,empirical analysis of the national railway operational mileage demon-strates that the weights determined by genetic algorithms exhibit superior forecasting performance in combination forecasting,show-casing higher accuracy and generalization capability compared to traditional methods.
关键词
遗传算法/组合预测/网格搜索/BP-神经网络Key words
genetic algorithm/combination forecasting/grid search/BP neural network引用本文复制引用
出版年
2024
NETL
NSTL
万方数据