小波分解结合神经网络的短期风电功率预测  

Short-term Wind Power Forecasting by Combining Wavelet Decomposition and Neural Network

作　　者：李保原 闫明晗 刘书圻 Li Baoyuan;Yan Minghan;Liu Shuqi

机构地区：[1]山东广域科技有限责任公司,山东东营257081

出　　处：《上海电气技术》2024年第4期17-20,共4页Journal of Shanghai Electric Technology

摘　　要：风电功率受风速等气象条件的影响,在短时间内波动很大。精准的风电功率预测可以减小风电波动性对电网的冲击,并可以帮助电力规划部门更好地制订电力调度计划,确保电力系统可靠性和稳定性。对此,融合信号处理和机器学习的优势,提出一种小波分解结合神经网络的短期风电功率预测方法。将风电功率视为信号,通过小波分解将输入转换为特征,通过神经网络提取特征,输出准确的预测。通过试验验证所提方法的有效性,并且所提方法的预测精度高于单纯前馈神经网络、循环神经网络、长短期记忆神经网络。将小波分解和神经网络有效结合,可以提高短期风电功率的预测精度。Wind power is influenced by meteorological conditions such as wind speed,and fluctuates significantly in a short time.Accurate forecasting of wind power can reduce the impact of wind power volatility on the power grid and assist power planning department in formulating better power dispatch plan,ensuring reliability and stability of the power system.To address this,a short-term wind power forecasting method combining wavelet decomposition and neural network was proposed,leveraging the advantages of both signal processing and machine learning.The wind power is treated as signal,the input is transformed into feature through wavelet decomposition,and the feature is extracted by using neural network to produce accurate forecasting.The effectiveness of the proposed method was verified through experiment,and the forecasting accuracy of the proposed method is higher than simple feedforward neural network,recurrent neural network,and long short-term memory neural network.The effective integration of wavelet decomposition and neural network can improve the forecasting accuracy of short-term wind power.

关 键 词：小波分解 神经网络 风力发电 功率 预测 

分 类 号：TM614[电气工程—电力系统及自动化]