针对风力机叶片在恶劣环境下易发生故障且检测困难的问题,提出一种新的故障诊断方法。该方法结合了倍频程理论与梅尔频率倒谱系数（MFCC）算法,通过引入对称可变倍频程技术对传统MFCC算法进行了优化。在频带划分上,依据叶片声音信号的特性与倍频程理论,重构物理频率与Mel频率之间的映射关系,以增强算法对分布在中频段与高频段的故障特征的提取能力,并有效减少噪声干扰。接着,利用K-均值聚类算法对优化后MFCC算法所提取的声学特征进行聚类分析,通过手肘法则确定叶片不同状态下的最佳聚类数,并根据短时能量分布去除噪声簇,实现了对不同状态叶片声音信号的有效区分。最后,基于随机森林算法构建分类器,对叶片故障进行准确诊断,验证了改进后的MFCC算法提取叶片声学特征及抗干扰的能力。
Abstract
Aiming at the problem that wind turbine blades are prone to fault and difficult to detect in harsh environments, a new fault diagnosis method is proposed. This method combines the Frequency octave theory and the Mel Frequency Cepstrum Coefficient （MFCC） algorithm, and optimizes the traditional MFCC algorithm by introducing the symmetric variable frequency octave-based technology. In terms of frequency band division, according to the characteristics of blade sound signals and octave theory, the mapping relationship between physical frequency and Mel frequency is reconstructed to enhance the algorithm’s ability to extract fault features distributed in the middle frequency band and high frequency band, and effectively reduce noise interference. Then, the K-means clustering algorithm is used to cluster the acoustic features extracted by the optimized MFCC algorithm. The elbow rule is used to determine the optimal number of clusters under different states of blades, and the noise clusters are removed according to the short-term energy distribution, so as to effectively distinguish the sound signals of different states of blades. Finally, a classifier based on random forest algorithm was constructed to accurately diagnose blade faults. It verifies the ability of the improved MFCC algorithm to extract the acoustic features of the wind turbine blades and anti-interference.
关键词
风力机叶片 /
故障诊断 /
聚类分析 /
声信号处理 /
梅尔频率倒谱系数 /
特征提取
Key words
wind turbine blades /
fault detection /
clustering algorithms /
acoustic signal processing /
MFCC /
feature extraction
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] SWENSON L, GAO L Y, HONG J R, et al.An efficacious model for predicting icing-induced energy loss for wind turbines[J]. Applied energy, 2022, 305: 117809.
[2] YANG K, RONGONG J A, WORDEN K.Damage detection in a laboratory wind turbine blade using techniques of ultrasonic NDT and SHM[J]. Strain, 2018, 54(6): e12290.
[3] PAN X, LIU Z D, XU R, et al.Early warning of damaged wind turbine blades using spatial-temporal spectral analysis of acoustic emission signals[J]. Journal of sound and vibration, 2022, 537: 117209.
[4] XU D, LIU P F, CHEN Z P.Damage mode identification and singular signal detection of composite wind turbine blade using acoustic emission[J]. Composite structures, 2021, 255: 112954.
[5] ZHOU W J, WANG Z J, ZHANG M S, et al.Wind turbine actual defects detection based on visible and infrared image fusion[J]. IEEE transactions on instrumentation and measurement, 2023, 72: 3509208.
[6] ALI FARHAN OGAILI A, ABDULHADY JABER A, HAMZAH M N. Wind turbine blades fault diagnosis based on vibration dataset analysis[J]. Data in brief, 2023, 49: 109414.
[7] LU Q, YE W X, YIN L F.Parallel multiple CNNs with temporal predictions for wind turbine blade cracking early fault detection[J]. IEEE transactions on instrumentation and measurement, 2024, 73: 3511511.
[8] CHEN B, YU S H, YU Y, et al.Acoustical damage detection of wind turbine blade using the improved incremental support vector data description[J]. Renewable energy, 2020, 156: 548-557.
[9] LIU H Q, ZHU W B, ZHOU Y, et al.Nonintrusive wind blade fault detection using a deep learning approach by exploring acoustic information[J]. The journal of the acoustical society of America, 2023, 153(1): 538.
[10] 张家安, 田家辉, 王铁成, 等. 基于改进MFCC算法的风力机叶片故障诊断方法[J]. 太阳能学报, 2024, 45(1): 285-290.
ZHANG J A, TIAN J H, WANG T C, et al.Wind turbine blade fault diagnosis method based on improved MFCC algorithm[J]. Acta energiae solaris sinica, 2024, 45(1): 285-290.
[11] 张鑫, 徐遵义, 何慧茹, 等. 基于RBM和SVM的风电机组叶片开裂故障预测[J]. 电力系统保护与控制, 2020, 48(15): 134-140.
ZHANG X, XU Z Y, HE H R, et al.Wind turbine blade cracking fault prediction based on RBM and SVM[J]. Power system protection and control, 2020, 48(15): 134-140.
[12] 金秋霞, 彭鹏, 孙萍玲, 等. 考虑噪声影响的风电场功率分配优化模型[J]. 太阳能学报, 2023, 44(4): 115-124.
JIN Q X, PENG P, SUN P L, et al.Optimization model of power distribution of wind farms considering noise impact[J]. Acta energiae solaris sinica, 2023, 44(4): 115-124.
[13] 吕晴. 倍频程分析新方法及其声功率检测应用研究[D]. 长沙: 湖南大学, 2018.
LYU Q.The research on a new method of octave analysis and its application in sound power detection[D]. Changsha: Hunan University, 2018.
[14] WANG L, ZHANG Z J, LUO X.A two-stage data-driven approach for image-based wind turbine blade crack inspections[J]. IEEE/ASME transactions on mechatronics, 2019, 24(3): 1271-1281.
[15] CHANTAMUNEE S, THAMRONGRAT P, THANATHAMATHEE P, et al.Unsupervised deep clustering with hard balanced constraint: application in disciplinary-focused student section formation[J]. IEEE access, 2024, 12: 98239-98253.
基金
河北省自然科学基金创新研究群体延续资助项目（E2024202298）
PDF(1798 KB)
