摘要
以云南省龙潭寨汛期与枯期输沙量时间序列预测为例,建立战争策略优化(WSO)算法、变色龙群算法(CSA)与极限学习机(ELM)相融合的组合模型。首先,在不同维度下选取4个基准函数对WSO、CSA进行仿真测试;其次,利用2层WPT将实例汛期与枯期输沙量时序数据分解为4个更具规律的子序列分量;最后,通过各分量训练样本构建ELM适应度函数,利用WSO、CSA对适应度函数进行寻优,利用寻优获得的最佳ELM超参数建立WPT-WSO-ELM、WPT-CSA-ELM模型对各子序列分量进行预测。将预测结果加和重构得到最终预测结果,并构建WPT-ELM模型及基于小波变换(WT)的WT-WSO-ELM、WT-CSA-ELM、WT-ELM模型作对比分析。对于基准函数及ELM适应度函数,WSO寻优效果优于CSA,具有较好的寻优精度及全局搜索能力;对汛期与枯期输沙量预测WPT-WSO-ELM模型预测精度优于WPT-CSA-ELM、WT-WSO-ELM、WT-CSA-ELM模型。
Taking the time series prediction of sediment discharge in Longtanzhai,Yunnan Province in flood season and dry season as an example,a combined model of war strategy optimization(WSO)algorithm,chameleon swarm algorithm(CSA)and extreme learning machine(ELM)is established.Firstly,four benchmark functions are selected to simulate and test WSO and CSA under different dimensional conditions.Secondly,the time-series data of sediment discharge in flood season and dry season are decomposed into four more regular sub-sequence components by using two-layer wavelet packet transform(WPT).Finally,the ELM fitness function is constructed through each component training sample and the WSO and CSA are used to optimize the fitness function,and the optimal ELM hyperparameters obtained by optimization is used to establish the WPT-WSO-ELM and WPT-CSA-ELM models to predict each sub-sequence components.The prediction results are added and reconstructed to obtain the final prediction results,and the WPT-ELM model and the WT-WSO-ELM,WT-CSA-ELM and WT-ELM models based on wavelet transform(WT)are constructed for comparative analysis.For the benchmark function and the ELM fitness function,the WSO optimization effect is better than that of the CSA,and has better optimization accuracy and global search ability.The prediction accuracy of the WPT-WSO-ELM model is better than that of the WPT-CSA-ELM,WT-WSO-ELM,WT-CSA-ELM models for the prediction of sediment discharge in flood season and dry season.
作者
许建伟
崔东文
XU Jianwei;CUI Dongwen(Yunnan Water Conservancy and Hydropower Survey and Design Institute,Kunming 650021,Yunnan,China;Yunnan Province Wenshan Water Bureau,Wenshan 663000,Yunnan,China)
出处
《水力发电》
CAS
2022年第11期36-42,共7页
Water Power
基金
云南省创新团队建设专项(2018HC024)
云南重点研发计划(科技入滇专项)
国家澜湄合作基金项目(2018-1177-02)。
关键词
输沙量预测
极限学习机
战争策略优化算法
变色龙群算法
小波包变换
仿真测试
sediment discharge prediction
extreme learning machine
war strategy optimization algorithms
chameleon swarm algorithm
wavelet packet transform
simulation test
