针对风电机组运行工况复杂,实际采集的振动信号存在分布差异,导致故障诊断模型的分类效果偏低问题,提出一种具有多核领域适应（MKDA）的多尺度卷积神经网络（MSCNN）风电机组轴承故障诊断研究方法（MKDA-MSCNN）。该方法通过迁移理论将已知风电机组知识迁移至目标风电机组实现故障诊断。首先,利用源域数据预训练MSCNN网络,再利用多核领域适应减小源域和目标域分布差异,最终获得目标风电机组故障诊断模型。试验结果表明,该文提出的MKDA-MSCNN方法在实际风电机组轴承故障诊断中分类精度高达96.17%,对比结果表明该文所提方法的故障分类准确度优于其他深度学习和深度迁移学习方法,对迁移学习理论在实际工程风电机组轴承故障诊断中的研究具有一定价值。
Abstract
Aiming at the problem that the classification effect of the fault diagnosis model is low due to the complexity of wind turbine operating conditions and the difference in the distribution of vibration signals actually collected, a multi-scale convolutional neural network（MSCNN） with multi-kernel domain adaptation（MKDA） was proposed in this paper（MKDA-MSCNN）. In this method, the known wind turbine knowledge was transferred to the target wind turbine through the transfer theory to achieve fault diagnosis. Firstly, the MSCNN model was pre-trained by source domain data, and the MKDA was used to reduce the distribution difference between the source domain and target domain, and finally, the target wind turbine fault diagnosis model was obtained. The test results show that the proposed MKDA-MSCNN method in actual wind turbines in bearing fault diagnosis classification accuracy is as high as 96.17%. The comparison results show that the fault classification accuracy of the proposed method is superior to the other deep learning and deep transfer learning methods, which is valuable for the study of transfer learning theory in the fault diagnosis of wind turbine bearings in practical engineering.
关键词
风电机组 /
故障诊断 /
滚动轴承 /
卷积神经网络 /
迁移学习
Key words
wind turbines /
fault diagnosis /
rolling bearing /
convolutional neural network /
transfer learning
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 苏连成, 邢美玲, 张慧. 基于组合预测模型的风电机组关键部位故障检测[J]. 太阳能学报, 2021, 42(10): 220-225.
SU L C, XING M L, ZHANG H.Fault detection of key components of wind turbine based on combination prediction model[J]. Acta energiae solaris sinica, 2021, 42(10): 220-225.
[2] 齐咏生, 刘飞, 李永亭, 等. 基于MK-MOMEDA和Teager能量算子的风电机组滚动轴承复合故障诊断[J]. 太阳能学报, 2021, 42(7): 297-307.
QI Y S, LIU F, LI Y T, et al.Compound fault diagnosis of wind turbine rolling bearing based on MK-MOMEDA and teager energy operator[J]. Acta energiae solaris sinica, 2021, 42(7): 297-307.
[3] 苗宝权, 陈长征, 罗园庆, 等. 基于自适应增强差分积形态滤波器的滚动轴承故障特征提取方法[J]. 机械工程学报, 2021, 57(9): 78-88.
MIAO B Q, CHEN C Z, LUO Y Q, et al.Rolling bearing fault feature extraction method based on adaptive enhanced differential product morphological filter[J]. Journal of mechanical engineering, 2021, 57(9): 78-88.
[4] XIANG L, SU H, LI Y.Research on extraction of compound fault characteristics for rolling bearings in wind turbines[J]. Entropy, 2020, 22(6): 682-697.
[5] CHEN B Y, SONG D L, ZHANG W H, et al.A performance enhanced time-varying morphological filtering method for bearing fault diagnosis[J]. Measurement, 2021, 176(10): 1-19.
[6] 向玲, 李营. 风电机组滚动轴承复合故障诊断研究[J]. 太阳能学报, 2021, 42(3): 90-97.
XING L, LI Y.Research on composite fault diagnosis of wind turbine rolling bearings[J]. Acta energiae solaris sinica, 2021, 42(3): 90-97.
[7] XU Z F, LI C, YANG Y.Fault diagnosis of rolling bearing of wind turbines based on the variational mode decomposition and deep convolutional neural networks[J]. Applied soft computing, 2020, 95: 106515.
[8] 张西宁, 刘书语, 余迪, 等. 改进深度卷积神经网络及其在变工况滚动轴承故障诊断中的应用[J]. 西安交通大学学报, 2021, 55(6): 1-8.
ZHANG X N, LIU S Y, YU D, et al.Improved deep convolutional neural network with applications to bearing fault diagnosis under variable conditions[J]. Journal of Xi’an Jiaotong University, 2021, 55(6): 1-8.
[9] WANG H, LIU Z L, PENG D D, et al.Understanding and learning discriminant features based on multi-attention 1DCNN for wheelset bearing fault diagnosis[J]. IEEE transactions on industrial informatics, 2019, 16(9): 5735-5745.
[10] XU M Q, WANG Y Q.An imbalanced fault diagnosis method for rolling bearing based on semi-supervised conditional generative adversarial network with spectral normalization[J]. IEEE access, 2021, 9: 27736-27747.
[11] LI Y T, JIANG W B, ZHANG G Y, et al.Wind turbine fault diagnosis based on transfer learning and convolutional autoencoder with small-scale data[J]. Renewable energy, 2021, 171(3): 67-78.
[12] SUN Y H, YU J B.Fault detection of rolling bearing using sparse representation-based adjacent signal difference[J]. IEEE transactions on instrumentation and measurement, 2020, 70: 1-16.
基金
国家自然科学基金（51675350; 51575361）
PDF(2114 KB)
