基于深度学习和广义S变换协同的风速预测
WIND SPEED PREDICTION SYNERGISTICALLY BASED ON DEEP LEARNING AND GENERALIZED S TRANSFORM
朱哲萱 1马汝为 1曹黎媛 1李春祥1
作者信息
- 1. 上海大学力学与工程科学学院,上海 200444
- 折叠
摘要
针对实测风速的非平稳性特点,提出一种基于深度学习和时频分析的风速混合预测方法.首先,采用经验模态分解(EMD)将风速分解为若干子层,由此得到趋势分量和脉动分量以降低风速的非线性.根据2个分量的时频特性,采用长短时记忆(LSTM)处理趋势分量,极限学习机(ELM)处理脉动分量.其次,引入广义S变换(GST)来获得预测过程中的时频特性.同时,采用改进的灰狼算法(IGWO)对GST、LSTM和ELM的参数进行优化.最后,以内蒙古某风场实测风速对所提模型进行验证,结果表明该模型具有较高的精度.
Abstract
A hybrid wind speed prediction model based on deep learning and time-frequency analysis is proposed,aiming at the non-stationary characteristics of wind speed.Firstly,empirical mode decomposition(EMD)is used to decompose the wind speed into several sub layers and summarized into a trend component and a fluctuating component to reduce the nonlinearity.According to the time-frequency characteristics of the two components,long short term memory(LSTM)is used to deal with the trend component while extreme learning machine(ELM)with the fluctuating component.Then,generalized S transform(GST)is innovatively introduced to obtain the time-frequency characteristics of the prediction process.Improved grey wolf algorithm(IGWO)is used to optimize the parameters of GST,LSTM and ELM at the same time.Finally,the proposed model is validated with the actual data of a wind farm in Inner Mongolia,and the results show that the model has accuracy.
关键词
风电场/风速/预测/长短时记忆/极限学习机/广义S变换Key words
wind farm/wind speed/prediction/long short-term memory/extreme learning machine/generalized S transform引用本文复制引用
出版年
2024
NETL
NSTL
万方数据