基于极限学习机的非平稳下击暴流风速预测
发布时间：2019-01-09 07:38
【摘要】：分别运用经验模态分解(empirical mode decomposition,EMD)法和快速集合经验模态分解(fast ensemble empirical mode decomposition,FEEMD)法将非平稳下击暴流风速分解为一系列稳态序列集,即固有模态分量.建立极限学习机(extreme learning machines,ELM)风速预测模型(EMD-ELM)和快速EMD-ELM(FEEMD-ELM),分别对分解后的非平稳脉动风速训练集和测试集进行预测.同时,将EMD和FEEMD与基于粒子群优化(particle swarm optimization,PSO)最小二乘支持向量机(least squares support vector machine,LSSVM)进行混合,形成EMD-PSO-LSSVM和FEEMD-PSO-LSSVM混合模型算法.通过比较这4种预测算法的结果发现,基于EMD-ELM和FEEMD-ELM的非平稳下击暴流风速预测模型更为准确高效,其中FEEMD-ELM模型预测最佳.
[Abstract]:The empirical mode decomposition (empirical mode decomposition,EMD) method and the fast set empirical mode decomposition (fast ensemble empirical mode decomposition,FEEMD) method are used to decompose the wind speed of non-stationary downburst flow into a series of steady state sequence sets, namely the intrinsic modal component. A wind speed prediction model (EMD-ELM) and a fast EMD-ELM (FEEMD-ELM) model for extreme learning machine (extreme learning machines,ELM) were established to predict the non-stationary pulsating wind speed training set and the test set, respectively. At the same time, the EMD and FEEMD are mixed with the least squares support vector machine (least squares support vector machine,LSSVM) based on particle swarm optimization (particle swarm optimization,PSO) to form the EMD-PSO-LSSVM and FEEMD-PSO-LSSVM hybrid model algorithm. By comparing the four prediction algorithms, it is found that the model based on EMD-ELM and FEEMD-ELM is more accurate and efficient, and the FEEMD-ELM model is the best.
【作者单位】： 上海大学土木工程系
【基金】：国家自然科学基金资助项目(51378304)
【分类号】：TP18
本文编号：2405319
[Abstract]:The empirical mode decomposition (empirical mode decomposition,EMD) method and the fast set empirical mode decomposition (fast ensemble empirical mode decomposition,FEEMD) method are used to decompose the wind speed of non-stationary downburst flow into a series of steady state sequence sets, namely the intrinsic modal component. A wind speed prediction model (EMD-ELM) and a fast EMD-ELM (FEEMD-ELM) model for extreme learning machine (extreme learning machines,ELM) were established to predict the non-stationary pulsating wind speed training set and the test set, respectively. At the same time, the EMD and FEEMD are mixed with the least squares support vector machine (least squares support vector machine,LSSVM) based on particle swarm optimization (particle swarm optimization,PSO) to form the EMD-PSO-LSSVM and FEEMD-PSO-LSSVM hybrid model algorithm. By comparing the four prediction algorithms, it is found that the model based on EMD-ELM and FEEMD-ELM is more accurate and efficient, and the FEEMD-ELM model is the best.
【作者单位】： 上海大学土木工程系
【基金】：国家自然科学基金资助项目(51378304)
【分类号】：TP18
【相似文献】
相关期刊论文 前1条
1 钟旺;李春祥;;基于极限学习机的非平稳下击暴流风速预测[J];上海大学学报(自然科学版);2018年03期
，本文编号：2405319
本文链接：https://www.wllwen.com/jianzhugongchenglunwen/2405319.html
