基于动态潜变量回归的风电机组叶片结冰监测研究

doi:10.3969/j.issn.2097-0706.2024.12.004

摘要/Abstract

摘要：

风电机组叶片结冰监测与预警对保障风电机组安全稳定运行具有重要意义。充分考虑结冰过程对风电机组性能累积的影响，提出一种符合机组动态运行特性的叶片结冰监测方法，从而提高风电机组叶片结冰监测准确率。通过孤立森林模型提取功率主带，为性能劣化模型的建立提供高质量的数据支撑。使用动态潜变量回归算法使质量变量在过程变量动态潜空间上的投影最大化，基于最小化回归误差提取风电机组动态运行过程输入变量和输出变量间潜在的结构化关系。采用向量自回归模型计算动态监测指标，若输出功率的监测指标越限则进行劣化告警。然后再使用孤立森林模型对劣化时段的叶尖速比和桨距角进行异常检测，若参数存在异常且环境温度小于0 ℃ 时发出叶片结冰预警。以我国西南某风场机组实际运行数据为例，验证了提出方法的有效性。

关键词: 叶片结冰监测, 故障预警, 孤立森林, 动态潜变量回归, 特征工程, 新型电力系统

Abstract:

The monitoring and warning of wind turbine blade icing is of great significance to ensure the safe and stable operation of wind turbines. Considering the accumulating effects of icing processes on the turbine performance， a wind turbine blade icing monitoring model is proposed that conforms to the dynamic operating characteristics of the unit，so as to improve the accuracy of wind turbine blade icing monitoring. The power main band is extracted by isolated forest algorithm to provide high quality data for the establishment of performance degradation model. A dynamic latent variable regression algorithm is used to maximize the projection of quality variable on the dynamic latent space， extract latent structured relations between process and quality variables in the dynamic operation of wind turbines based on minimizing regression errors. The vector auto regression model is used to calculate the dynamic monitoring indicators， and if the monitoring indicators of the output power exceed the limit， the warning of deterioration is carried out. Then the tip speed ratio and pitch angle of the degradation data are abnormally detected based on isolated forest and if the parameters are abnormal and the ambient temperature is below 0 ℃， a blade icing warning will be issued. The actual operation data of a wind turbine in southwest China is used as an example to verify the effectiveness of the method in this paper.

Key words: blade icing monitoring, fault alert, isolated forest, dynamic latent variable regression, feature engineering, new power system

中图分类号:

TK83

肖碧涛, 刘宇, 赖晓路. 基于动态潜变量回归的风电机组叶片结冰监测研究[J]. 综合智慧能源, 2024, 46(12): 29-35.

XIAO Bitao, LIU Yu, LAI Xiaolu. Wind turbine blades icing monitoring based on dynamic latent variable regression[J]. Integrated Intelligent Energy, 2024, 46(12): 29-35.

图/表 8

图1

图2

图3

图 4

图 5

表1

图 6

图 7

参考文献 16

[1]	董健, 柳亦兵, 滕伟, 等. 基于BP-Adaboost算法的风电机组叶片结冰检测[J]. 可再生能源, 2021, 39(5):632-636.
	DONG Jian, LIU Yibing, TENG Wei, et al. Wind turbine blade ice detection based on BP-Adaboost algorithm[J]. Renewable Energy Resources, 2021, 39(5):632-636.
[2]	RIZK P, SALEH N, YOUNES R, et al. Hyperspectral imaging applied for the detection of wind turbine blade damage and icing[J]. Remote Sensing Applications:Society and Environment, 2020,18:100291.
[3]	MUNOZ C, MARQUEZ F, TOMAS J. Ice detection using thermal infrared radiometry on wind turbine blades[J]. Measurement, 2016,93:157-163.
[4]	SHOJA S, BERBYUK V, BOSTROM A. Guided wave-based approach for ice detection on wind turbine blades[J]. Wind Engineering, 2018, 42(5):483-495.
[5]	PATRICE R, JEAN L, RJEAN, et al. A new atmospheric icing detector based on thermally heated cylindrical probes for wind turbine applications[J]. Cold Regions Science and Technology, 2018,148:131-141.
[6]	郭奕兵. 基于FBG传感和石墨烯薄膜的风机结冰监测与除冰[D]. 哈尔滨: 哈尔滨工业大学, 2018.
	GUO Yibing. Graphene film combined with FBG sensor for icing monitoring and de-icing of wind turbine[D]. Harbin: Harbin Institute of Technology, 2018.
[7]	李宁波, 闫涛, 李乃鹏, 等. 基于SCADA数据的风机叶片结冰检测方法[J]. 发电技术, 2018, 39(1):58-62. doi: 10.12096/j.2096-4528.pgt.2018.010
	LI Ningbo, YAN Tao, LI Naipeng, et al. Ice detection method by using SCADA data on wind turbine blades[J]. Power Generation Technology, 2018, 39(1):58-62. doi: 10.12096/j.2096-4528.pgt.2018.010
[8]	范大千, 刘博嵩, 郭鹏. 基于AdaBoost算法的多参数模型风电机组叶片结冰监测与预警研究[J]. 华电技术, 2021, 43(8):20-26.
	FAN Daqian, LIU Bosong, GUO Peng. Wind turbine blades icing detection with multi-parameter models based on AdaBoost algorithm[J]. Huadian Technology, 2021, 43(8):20-26.
[9]	XIAO J, LI C Y, LIU B, et al. Prediction of wind turbine blade icing fault based on selective deep ensemble model[J]. Knowledge-Based Systems, 2022,242:108290.
[10]	TAO T, LIU Y Q, QIAO Y H, et al. Wind turbine blade icing diagnosis using hybrid features and Stacked-XGBoost algorithm[J]. Renewable Energy. 2021,180:1004-1013.
[11]	李大中, 王超, 李颖宇. 基于XGBoost算法的风机叶片结冰状态评测[J]. 电力科学与工程, 2019, 35(9):43-48. doi: 1672-0792(2019)09-0043-06
	LI Dazhong, WANG Chao, LI Yingyu. Evaluation of fan blade icing based on XGBoost algorithm[J]. Electric Power Science and Engineering, 2019, 35(9):43-48. doi: 1672-0792(2019)09-0043-06
[12]	王金轩, 汤占军, 詹跃东, 等. 基于卷积神经网络的风机叶片结冰故障检测[J]. 计算机仿真, 2021, 38(12):85-88.
	WANG Jinxuan, TANG Zhanjun, ZHAN Yuedong, et al. Fan blade icing fault detection based on convolution neural network[J]. Computer Simulation, 2021, 38(12):85-88.
[13]	LI F Y, CUI H M, SU H J, et al. Icing condition prediction of wind turbine blade by using artificial neural network based on modal frequency[J]. Cold Regions Science and Technology, 2022,194:103467.
[14]	CHEN T Q, GUESTRIN C. XGBoost:A scalable tree boosting system[C]// Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining.San Francisco,2016:785-794.
[15]	LESOUPLE J, BAUDOIN C, SPIGAI M, et al. Generalized isolationforest for anomaly detection[J]. Pattern Recognition Letters, 2021,149:109-119.
[16]	ZHU Q Q, QIN S J. Latent variable regression for process and quality modeling[C]// 2019 1st International Conference on Industrial Artificial Intelligence.Shenyang,2019: 1-6.