%0 Journal Article
%T 基于新的组合模型的短期风速预测<br>Short-Term Wind Speed Prediction Based on a New Combined Model
%A 胡文博
%A 袁柳洋
%A 马建清
%J Operations Research and Fuzziology
%P 1120-1129
%@ 2163-1530
%D 2024
%I Hans Publishing
%R 10.12677/ORF.2024.141103
%X 鉴于风速序列呈现高度随机和不稳定的特性，本文构建了一个新的组合模型用于短期风速预测，该模型由粒子群优化–时变滤波–经验模态分解(PSO-TVF-EMD)、排列熵(PE)、长短期记忆网络(LSTM)和自回归差分移动平均(ARIMA)组成。首先利用PSO-TVF-EMD算法将原始风速序列分解为若干模态分量，以简化风速序列的复杂性；其次，使用PE把风速子模态分为高频序列和低频序列，并对这两种序列分别构建LSTM和ARIMA预测模型；最终，将子序列预测结果叠加，得出最终的风速预测值。试验结果表明，新的组合预测模型一定程度上增加了预测的精度。<br />
In view of the highly random and unstable characteristics of wind speed series, a new combined model is constructed for short-term wind speed prediction. The model consists of particle swarm optimization-time-varying filtering-empirical mode decomposition (PSO-TVF-EMD), permutation entropy (PE), long-short term memory network (LSTM) and auto regressive integrated moving average (ARIMA). Firstly, the original wind speed series are decomposed into several modal components by using PSO-TVF-EMD algorithm to simplify the complexity of the wind speed series, and secondly, the wind speed sub-modes are divided into high frequency series and low frequency series by using PE algorithm, the LSTM and ARIMA prediction models are constructed for the two series, and the wind speed prediction values are obtained by superim-posing the results of the sub<span>-</span>series. The experimental results show that the new combined forecasting model increases the forecasting precision to some extent.
%K 时变滤波，粒子群优化，排列熵模型，长短期记忆网络，短期风速预测<br>Time-Varying Filtering
%K Particle Swarm Optimization
%K Permutation Entropy Model
%K Long-Term and Short-Term Memory Network
%K Short-Term Wind Speed Prediction
%U http://www.hanspub.org/journal/PaperInformation.aspx?PaperID=82310