基于两阶段特征选择和改进NRBO-BiLSTM的短期风电功率预测方法

袁博; 周霞; 戴剑丰; 张腾飞

doi:10.16516/j.ceec.2025-224

欢迎来到能源电力期刊网！

English Version 期刊登录

您当前的位置：

首页 >

文章列表页 >

基于两阶段特征选择和改进NRBO-BiLSTM的短期风电功率预测方法

极端气候事件下的能源与交通韧性 | 更新时间：2026-10-14

- 基于两阶段特征选择和改进NRBO-BiLSTM的短期风电功率预测方法
- A Short-Term Wind Power Prediction Method Based on Two-Stage Feature Selection and Improved NRBO-BiLSTM
- 南方能源建设 2026年13卷第3期页码：138-151
- 作者机构：
  
  南京邮电大学自动化学院/人工智能学院,江苏,南京,210023
- 作者简介：
  
  袁博，ORCID: 0009-0004-1578-321X2000-，男，硕士在读，从事多能源联合调频研究工作（e-mail）577990235@qq.com
  周霞，（e-mail）zhouxia@njupt.edu.cn
  戴剑丰，1989-，男，讲师，博士，从事新能源发电并网控制及其对电力系统的影响工作（e-mail）daijianfeng@njupt.edu.cn
  张腾飞，1980-，男，教授，博士，从事智能配电网与配电物联网技术研究工作（e-mail）zhangtf@njupt.edu.cn
- 基金信息：
  
  国家自然科学基金项目“泛在电力物联配电网需求侧响应驱动的多粒度聚合与负荷管控”（62073173）
- DOI：10.16516/j.ceec.2025-224
  中图分类号： TK89;TM614;TP18
- 纸质出版：2026
- 稿件说明：
移动端阅览
袁博, 周霞, 戴剑丰, 等. 基于两阶段特征选择和改进NRBO-BiLSTM的短期风电功率预测方法[J]. 南方能源建设, 2026, 13(3): 138-151.

.
袁博, 周霞, 戴剑丰, 等. 基于两阶段特征选择和改进NRBO-BiLSTM的短期风电功率预测方法[J]. 南方能源建设, 2026, 13(3): 138-151. DOI： 10.16516/j.ceec.2025-224.

. DOI： 10.16516/j.ceec.2025-224.

摘要

目的

针对风电功率预测中特征因素太多、关键特征与功率间映射关系难以有效挖掘和预测精度不高的问题，提出一种考虑两阶段特征选择的短期风电功率预测方法。

方法

首先，采用基于互信息（Mutual Information，MI）的最大相关-最小冗余判据（minimal Redundancy Maximal Relevance，mRMR）对风电功率预测数据中的风速、风向等特征进行相关性分析，通过最大化特征与风电功率之间的相关性并最小化特征间的冗余度，实现特征的初步筛选；接着，利用随机森林（Random Forest，RF）算法中的袋外（Out-of-Bag，OOB）重要性评价方法对筛选后的特征进行重要性评价，进行特征的二次筛选以获取最优特征子集；然后，对牛顿-拉夫逊优化（Newton-Raphson Based Optimizer，NRBO）算法的自适应参数进行改进，并利用改进后的NRBO优化算法对双向长短期记忆（Bidirectional Long Short-Term Memory，BiLSTM）网络的相关参数进行优化；最后将最优特征子集输入到NRBO-BiLSTM风电功率预测模型中进行短期风电功率预测。

结果

算例分析表明，所提两阶段特征选择方法能有效剔除特征间冗余及低重要性特征，最优特征子集显著提升了模型性能；改进NRBO算法优化效果显著，预测精度显著优于对比模型。

结论

文章成功地将两阶段特征选择与改进的NRBO算法相结合，所提模型能够精准捕捉风电功率的波动规律，有效提升了短期风电功率的预测精度。

Abstract

Objective

To address the challenges of excessive feature factors

difficulties in effectively mining the mapping relationship between key features and power

and low prediction accuracy in wind power forecasting

this paper proposes a short-term

wind power prediction method that incorporates two-stage feature selection.

Method

First

the maximum relevance minimum redundancy (mRMR) criterion based on mutual information (MI) was used to analyze the correlation among features such as wind speed and direction in wind power prediction data. Initial feature screening was achieved by maximizing the correlation between features and wind power while minimizing inter-feature redundancy. Next

the out-of-bag (OOB) feature importance evaluation method within the random forest (RF) algorithm was employed to assess the significance of the pre-screened features

and a secondary screening was performed to obtain the optimal feature subset. Then

the adaptive parameters of the Newton-Raphson based optimizer (NRBO) were improved

and the enhanced NRBO algorithm was used to optimize the parameters of a bidirectional long short-term memory network (BiLSTM). Finally

the optimal feature subset was input into the NRBO-BiLSTM model to predict short-term wind power.

Result

The case study analysis shows that the proposed two-stage feature selection method can effectively eliminate the redundancy among features and low-importance features

and the optimal feature subset significantly improves the model performance; The improved NRBO algorithm has a remarkable optimization effect

and its prediction accuracy is significantly better than that of the comparison models.

Conclusion

This study successfully combines two-stage feature selection with an improved NRBO optimization algorithm. The proposed model can accurately capture the fluctuation patterns of wind power and effectively improve the prediction accuracy of short-term wind power.

关键词

Keywords

references

WU W L, YANG J, QI H, et al. Analysis of extreme climate characteristics and risks of "wind and solar" in Ningxia new energy concentration area [J]. Power systems and big data, 2025, 28(9): 30-37. DOI: 10.19317/j.cnki.1008-083x.2025.09.004.

LI B J, FU X B, SHI K P, et al. Research on ultra short term combination prediction of wind power based on adversarial neural network [J]. Power electronics, 2025, 59(10): 89-95. DOI: 10.20222/j.cnki.cn61-1124/tm.20250704.008.

YU J, YIN X G, LAI J M, et al. Adaptive leg-balancing control strategy of an MMC in an asymmetric AC grid fault [J]. Power system protection and control, 2024, 52(1): 1-12. DOI: 10.19783/j.cnki.pspc.230122.

ZHENG T T , ZHAO X M, ZHENG L F, et al. Reactive power characteristic analysis and transient support of doubly-fed induction under power grid fault [J]. Power electronics, 2024, 58(1): 77-80. DOI: 10.3969/j.issn.1000-100X.2024.01.019.

WU Y H, WANG Y S, XU H, et al. Survey of wind power output power forecasting technology [J]. Journal of frontiers of computer science technology, 2022, 16(12): 2653-2677. DOI: 10.3778/j.issn.1673-9418.2205028.

LIU J P, DENG J M, GAO P Y, et al. Ultra short term wind power prediction based on VMD-IMPA-SVM [J]. Smart power, 2024, 52(7): 24-31, 79. DOI: 10.3969/j.issn.1673-7598.2024.07.004.

HODGE B M, MARTINEZ-ANIDO C B, WANG Q, et al. The combined value of wind and solar power forecasting improvements and electricity storage [J]. Applied energy, 2018, 214: 1-15. DOI: 10.1016/j.apenergy.2017.12.120.

HAO Y, TIAN C S. A novel two-stage forecasting model based on error factor and ensemble method for multi-step wind power forecasting [J]. Applied energy, 2019, 238: 368-383. DOI: 10.1016/j.apenergy.2019.01.063.

DENG W S, CHE J F, WANG M Q, et al. Day-ahead power forecasting method of wind cluster based on grid numerical weather prediction [J]. Southern power system technology, 2024, 18(6): 51-57, 78. DOI: 10.13648/j.cnki.issn1674-0629.2024.06.007.

DAUT M A M, HASSAN M Y, ABDULLAH H, et al. Building electrical energy consumption forecasting analysis using conventional and artificial intelligence methods: a review [J]. Renewable and sustainable energy reviews, 2017, 70: 1108-1118. DOI: 10.1016/j.rser.2016.12.015.

LI R J, LI L X. Short-term wind power prediction based on VMD-CNN-LSTM model [J]. Journal of Shenyang Institute of Engineering (natural science edition), 2024, 20(1): 6-13. DOI: 10.13888/j.cnki.jsie(ns).2024.01.002.

LIU Y J, FAN Y F, BAI X Y, et al. Short-term wind power prediction based on feature crossover mechanism and error compensation [J]. Transactions of China electrotechnical society, 2023, 38(12): 3277-3288. DOI: 10.19595/j.cnki.1000-6753.tces.220477.

JIANG M N, HE Y, ZHANG T Q, et al. Wind power prediction based on feature selection and error correction [J]. Distributed energy, 2023, 8(2): 37-43. DOI: 10.16513/j.2096-2185.DE.2308205.

LUO T M, SONG J W, DUAN J L, et al. Ultra short-term prediction of offshore wind power based on CNN-LSTM [J]. Electrical equipment and economy, 2024(6): 350-353. DOI: 10.3969/j.issn.1673-8845.2024.06.113.

XU H L. Adaptive feature selection-based ultra-short-term wind farm cluster power forecasting using optimized graph convolutional neural network [D]. Baoding: North China Electric Power University, 2024. DOI: 10.27139/d.cnki.ghbdu.2024.000090.

FU B Z, WANG W, REN G R, et al. Short-term wind power prediction with deep convolutional-temporal networks considering enhanced feature selection [J]. Journal of Chinese society of power engineering, 2024, 44(10): 1565-1573. DOI: 10.19805/j.cnki.jcspe.2024.230510.

JI L P, HUANG J T, LI Y F, et al. A wind power prediction feature selection method based on multi-information fusion [J]. Journal of Lanzhou University of Technology, 2024, 50(5): 94-100. DOI: 10.3969/j.issn.1673-5196.2024.05.013.

LUO X Y, LIU J, YANG B, et al. Ultra short-term wind power prediction based on improved ospery optimization algorithm and VMD-LSTM [J]. Acta energiae solaris sinica, 2025, 46(3): 652-660. DOI: 10.19912/j.0254-0096.tynxb.2023-1942.

WANG D, FU X J, DU S Q. Ultra-short-term wind power prediction based on adaptive VMD-LSTM [J]. Journal of Nanjing University of Information Science Technology, 2025, 17(1): 74-87. DOI: 10.13878/j.cnki.jnuist.20240404002.

LU Q X, MA G, TU M F. Ultra-short-term wind power prediction based on IEWT-FE-BO-LSTM model [J]. Water resources and power, 2023, 41(1): 217-220. DOI: 10.20040/j.cnki.1000-7709.2023.20220494.

MEI R, TANG Y M, GU W, et al. Model of offshore wind power forecast considering the variation of wind speed in second-level time scale [C]// Anon. Proceedings of 2021 IEEE Sustainable Power and Energy Conference, Nanjing, China, December 23-25, 2021. Nanjing: IEEE, 2021: 3659-3663. DOI: 10.1109/iSPEC53008.2021.9735605.

WU Y B, ZHANG J Z, DENG F J, et al. Anomaly data identification of wind power in wind farm with the criterion of variance change rate and quartile [J]. Electric power engineering technology, 2023, 42(4): 141-148. DOI: 10.12158/j.2096-3203.2023.04.015.

WANG F. Anomaly detection of bladeIcing in wind turbines based on variational autoencoder and adversarial SVDD algorithm [J]. Power systems and big data, 2025, 28(11): 54-61. DOI: 10.19317/j.cnki.1008-083x.2025.11.007.

CONG Y, LIU H Q, WANG Y Y, et al. Optimization and test verification of active power aggregation control for multiple wind turbines [J]. Inner Mongolia electric power, 2026, 44(2): 19-26. DOI: 10.19929/j.cnki.nmgdljs.2026.0017.

LI C Q. Wind farm power prediction based on feature selection and CEEMDAN-CNN-LSTM model [J]. Shenyang: Shenyang Agricultural University, 2024. DOI: 10.27327/d.cnki.gshnu.2024.001580.

BREIMAN L. Random forests [J]. Machine learning, 2001, 45(1): 5-32. DOI: 10.1023/A:1010933404324.

HOCHREITER S, SCHMIDHUBER J. LSTM can solve hard long time lag problems [C]// Anon. Proceedings of the 10th International Conference on Neural Information Processing Systems, Denver, Colorado, December 3-5, 1996. Denver: MIT Press, 1996: 473-479.

SOWMYA R, PREMKUMAR M, JANGIR P. Newton-Raphson-based optimizer: a new population-based metaheuristic algorithm for continuous optimization problems [J]. Engineering applications of artificial intelligence, 2024, 128: 107532. DOI: 10.1016/j.engappai.2023.107532.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

储能技术在发电侧的综合应用与进展

深地储气与压缩空气储能数值模拟研究进展

基于电动汽车有序充电的光储充电站双层优化配置

面向火电调频的双电层电容器储能系统设计与特性试验

基于切圆布气的人工硐室壁温抑制技术