| 引用本文: | 王康德,刘文泽,陈泽,黄展鸿,余涛,潘振宁.基于运行状态与功率特性引导的覆冰天气下风电机组功率预测[J].电力自动化设备,2024,44(11):88-93，133. |
| WANG Kangde,LIU Wenze,CHEN Ze,HUANG Zhanhong,YU Tao,PAN Zhenning.Power prediction of wind turbine under icing weather based on operation state and power characteristic guidance[J].Electric Power Automation Equipment,2024,44(11):88-93，133. |
|
| 本文已被：浏览 7220次 下载 1142次 |
码上扫一扫！ |
|
|
| 基于运行状态与功率特性引导的覆冰天气下风电机组功率预测 |
|
王康德1, 刘文泽1, 陈泽1, 黄展鸿1,2, 余涛1,2, 潘振宁1,2
|
|
1.华南理工大学 电力学院，广东 广州 510641;2.广东省电网智能量测与先进计量企业重点实验室，广东 广州 510640
|
|
| 摘要: |
| 针对覆冰天气下风电机组功率预测精度不足的问题，提出一种基于运行状态与功率特性引导的覆冰天气下风电机组功率预测方法。针对覆冰天气下风电机组样本不平衡的问题，采用贝叶斯优化的改进轻量级梯度提升机模型对覆冰天气下的风电机组数据样本进行状态判别并建立运行状态特征标签；针对覆冰天气下非停机运行小样本的问题，构建基于功率特性引导的风电机组功率预测模型，降低功率预测模型对数据的依赖，提升在覆冰天气运行场景下的预测准确性。采用广西某风电场风电机组数据进行算例验证，结果表明，相较于典型预测方法，所提方法在覆冰天气下的预测性能显著提升。 |
| 关键词: 覆冰天气 不平衡样本 小样本学习 轻量级梯度提升机 风电机组功率特性 风电 |
| DOI：10.16081/j.epae.202408002 |
| 分类号:TM614 |
| 基金项目:国家自然科学基金资助项目(52207105) |
|
| Power prediction of wind turbine under icing weather based on operation state and power characteristic guidance |
|
WANG Kangde1, LIU Wenze1, CHEN Ze1, HUANG Zhanhong1,2, YU Tao1,2, PAN Zhenning1,2
|
|
1.School of Electric Power, South China University of Technology, Guangzhou 510641, China;2.Guangdong Provincial Key Laboratory of Intelligent Measurement and Advanced Metering of Power Grid, Guangzhou 510640, China
|
| Abstract: |
| Aiming at the problem of insufficient power prediction accuracy of wind turbine under icing weather, a power prediction method for wind turbine under icing weather is proposed based on the guidance of operation state and power characteristic. Aiming at the sample imbalance problem of wind turbine under icing weather, Bayesian optimization-improved light gradient boosting machine model is adopted to discriminate the state of wind turbine data samples under icing weather and the operation state feature labels are established. Aiming at the small sample problem of non-stop operation under icing weather, a power prediction model for wind turbine is constructed based on the power characteristic guidance, which reduces the dependence of power prediction model on the data and improves the prediction accuracy under the icing weather operation scenario. The wind turbine data of a wind farm in Guangxi is used for example verification, and the results show that the prediction performance of the proposed method is significantly improved under icing weather compared with the typical prediction methods. |
| Key words: icing weather imbalance sample small sample learning light gradient boosting machine power characteristic of wind turbine wind power |
