基于改进去噪扩散概率模型的风电机组故障样本生成方法

PDF(2340 KB)

振动与冲击 ›› 2025, Vol. 44 ›› Issue (4) : 286-297.

故障诊断分析

基于改进去噪扩散概率模型的风电机组故障样本生成方法

孟昱煜1，张沣琦1，火久元*1，2，常琛1，陈峰1

作者信息 +

Wind turbine fault sample generation method based on improved denoising diffusion probability model

MENG Yuyu1, ZHANG Fengqi1, HUO Jiuyuan*1,2, CHANG Chen1, CHEN Feng1

Author information +

文章历史 +

摘要

为解决风电机组故障诊断中故障样本不足而导致模型准确率不高的问题，将当下备受关注的数据增强方法-去噪扩散概率模型(DDPM)引入到故障诊断领域以生成大量高质量的故障样本数据集。因此，结合Transformer网络，提出了一种DDPM-Transformer风电机组故障样本生成方法。首先，将用于计算机视觉图像生成领域的DDPM模型应用于风电机组故障诊断领域中，通过前向加噪过程将数据逐渐转化为噪声，再通过逆向去噪过程将噪声逐步恢复为原始数据，实现从噪声中生成故障数据，解决数据不平衡问题；其次，通过对原始DDPM中使用的U-net模块进行改进，使用Transformer模型替换U-net网络，利用扩散后的数据和添加的噪声训练Transformer模型，实现噪声预测，以提高故障数据的生成质量；最后，使用多种生成模型评价指标对生成的故障数据进行评价，在SCADA故障数据生成中论证改进DDPM-Transformer模型的性能。通过实验证明，所提DDPM-Transformer模型与现有的生成模型相比，MMD最大提升0.13，PSNR最大提升7.8。所提模型可以有效地生成质量更高的风电机组故障样本，从而基于该样本集辅助训练基于深度学习的故障诊断模型，可以使诊断模型具有更高精度和良好的稳定性。

Abstract

In order to solve the problem that the model accuracy is not high due to insufficient fault samples in the fault diagnosis of wind turbine, the current data enhancement method, denoising diffusion probability model (DDPM), is introduced into the fault diagnosis field to generate a large number of high-quality fault sample data sets. Therefore, with Transformer network, a DDPM-Transformer wind turbine fault samples generation method was proposed. Firstly, the DDPM model used in the field of computer vision image generation is applied to the field of wind turbine fault diagnosis, where the data is gradually transformed into noise through the forward noise addition process, and then the noise is progressively restored to the original data through the inverse denoising process, so as to realize the generation of fault data from noise and solve the problem of data imbalance. Secondly, by improving the U-net module used in the original DDPM and replacing the U-net network with the Transformer model, the noise prediction is realized by training the Transformer model using the diffused data and the noise improve the quality of fault data generation. Finally, a variety of generation model evaluation indexes are used to evaluate the generated fault data, and the improved performance of DDPM-Transformer model is demonstrated in SCADA fault data generation. Experiments show that compared with existing generation models, the proposed DDPM-Transformer model has a maximum increase of 0.13 MMD and 7.8 PSNR. The model in this paper can effectively generate higher-quality wind turbine fault samples, so that the fault diagnosis model based on deep learning can be trained based on the sample set, which can make the diagnosis model have higher accuracy and good stability.

导出引用

孟昱煜1, 张沣琦1, 火久元1, 2, 常琛1, 陈峰1. 基于改进去噪扩散概率模型的风电机组故障样本生成方法[J]. 振动与冲击, 2025, 44(4): 286-297

MENG Yuyu1, ZHANG Fengqi1, HUO Jiuyuan1, 2, CHANG Chen1, CHEN Feng1. Wind turbine fault sample generation method based on improved denoising diffusion probability model[J]. Journal of Vibration and Shock, 2025, 44(4): 286-297

参考文献

[1] Seeking RJ, Verma AS, Li Y, Teuwen J, Jiang Z.Offshore wind turbine operations and maintenance:A state-of-the-art review.Renewable and Sustainable Energy Reviews ，2021;144(1):110886.
[2] Qiao W, Lu D. A survey on wind turbine condition monitoring and fault diagnosis–Part I:components and subsystems.IEEE Transactions on Industrial Electronics， 2015;62(10):6536–45.
[3] Rezamand M, Kordestani M, Carriveau R, Ting DS, Orchard ME, Saif M. Critical wind turbine components prognostics: A comprehensive review. IEEE Transactions on Instrumentation and Measurement ，2020;69(12):9306–28.
[4] Tang H H, Zhang K, Wang B, et al. Early bearing fault diagnosis for imbalanced data in offshore wind turbine using improved deep learning based on scaled minimum unscented kalman filter[J]. Ocean Engineering, 2024, 300: 117392.
[5] Barua S ,Islam M M ,0001 Y X , et al.MWMOTE-Majority Weighted Minority Oversampling Technique for Imbalanced Data Set Learning.[J].IEEE Transactions on Knowledge and Data Engineering ,2014,26(2):405-425.
[6] Chatterjee S, Byun Y C. Highly imbalanced fault classification of wind turbines using data resampling and hybrid ensemble method approach[J]. Engineering Applications of Artificial Intelligence, 2023, 126: 107104.
[7] Tang M, Meng C, Wu H, et al. Fault detection for wind turbine blade bolts based on GSG combined with CS-LightGBM[J]. Sensors, 2022, 22(18): 6763.
[8] Yi H, Jiang Q, Yan X, et al. Imbalanced classification based on minority clustering synthetic minority oversampling technique with wind turbine fault detection application[J]. IEEE Transactions on Industrial Informatics, 2020, 17(9): 5867-5875.
[9] Yang Z, Xu M, Wang S, et al. Detection of wind turbine blade abnormalities through a deep learning model integrating VAE and neural ODE[J]. Ocean Engineering, 2024, 302: 117689.
[10] Jia Z, Yu B. A fault diagnosis method for rolling bearings of wind turbine generators based on MCGAN data enhancement[J]. SN Applied Sciences, 2023, 5(10): 259.
[11] Yang L, Zhang Z. Wind turbine gearbox failure detection based on SCADA data: A deep learning-based approach[J]. IEEE Transactions on Instrumentation and Measurement, 2020, 70: 1-11..
[12] Jiarui L,Guotian Y,Xinli L, et al. Wind turbine anomaly detection based on SCADA: A deep autoencoder enhanced by fault instances.[J]. ISA transactions,2023,139.
[13] Chen M, Shao H, Dou H, et al. Data augmentation and intelligent fault diagnosis of planetary gearbox using ILoFGAN under extremely limited samples[J]. IEEE Transactions on Reliability, 2022.
[14] Dosovitskiy A, Brox T. Generating images with perceptual similarity metrics based on deep networks[J]. Advances in neural information processing systems, 2016, 29.
[15] Sohl-Dickstein J, Weiss E, Maheswaranathan N, et al. Deep unsupervised learning using nonequilibrium thermodynamics[C], International conference on machine learning. PMLR, 2015: 2256-2265.
[16] Ho J, Saharia C, Chan W, et al. Cascaded diffusion models for high fidelity image generation[J]. The Journal of Machine Learning Research, 2022, 23(1): 2249-2281.
[17] Khrulkov V, Ryzhakov G, Chertkov A, et al. Understanding ddpm latent codes through optimal transport[J]. arxiv preprint arxiv:2202.07477, 2022..
[18] Li H, Yang Y, Chang M, et al. Srdiff: Single image super-resolution with diffusion probabilistic models[J]. Neurocomputing, 2022, 479: 47-59.
[19] Khader F, Müller-Franzes G, Tayebi Arasteh S, et al. Denoising diffusion probabilistic models for 3D medical image generation[J]. Scientific Reports, 2023, 13(1): 7303.
[20] Nai C, Pan B, Chen X, et al. Reliable precipitation nowcasting using probabilistic diffusion models[J]. Environmental Research Letters, 2024.
[21] Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models, 2020.
[22] Dosovitskiy A, Beyer L, Kolesnikov A, et al. An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale[J]. 2020.C. DOERSCH , Tutorial on variational autoencoders, Stat 1050 (2016) 13 .
[23] Li Q, Yao N, Zhao J, et al. Self attention mechanism of bidirectional information enhancement[J]. Applied Intelligence, 2022: 1-9.
[24] Senapati P, Basu A, Deb M, et al. Sharp dense U-Net: an enhanced dense U-Net architecture for nucleus segmentation[J]. International Journal of Machine Learning and Cybernetics, 2023: 1-16.
[25] Xu D, Tan J C, Hsu C J, et al. Denoising diffusion probabilistic models to predict the density of molecular clouds[J]. The Astrophysical Journal, 2023, 950(2): 146.
[26] Wu Y, He K. Group normalization[C]Proceedings of the European conference on computer vision (ECCV). 2018: 3-19.
[27] 火久元,张沣琦,孟昱煜,等.山西晋城泽州风电场10台风电机组故障数据集（2021-2022年）[DB/OL].国家冰川冻土沙漠科学数据中心(http://www.ncdc.ac.cn), 2024. https://cstr.cn/CSTR:11738.11.NCDC.NIEER.DB6400.2024.
HUO Jiu-yuan, Zhang Feng-qi, Meng Yu-yu,et al. Shanxi Jincheng Zizhou Wind Farm 10 Wind Turbine Failure Dataset（2021-2022）[DB/OL]. National Cryosphere Desert Data Center(http://www.ncdc.ac.cn), 2024. https://cstr.cn/CSTR:11738.11.NCDC.NIEER.DB6400.2024.
[28] Zhang X, He L, Wang X, et al. Transfer fault diagnosis based on local maximum mean difference and K-means[J]. Computers & Industrial Engineering, 2022, 172: 108568.
[29] Lyu Q, Zhao N, Yang Y, et al. A diffusion probabilistic model for traditional Chinese landscape painting super-resolution[J]. Heritage Science, 2024, 12(1): 4.
[30] Zhao X, Jia K. Cloud removal in remote sensing using sequential-based diffusion models[J]. Remote Sensing, 2023, 15(11): 2861.
[31] Perera M V, Nair N G, Bandara W G C, et al. SAR despeckling using a denoising diffusion probabilistic model[J]. IEEE Geoscience and Remote Sensing Letters, 2023.
[32] Du H, Xu J, Du Z, et al. MF-MNER: Multi-models Fusion for MNER in Chinese Clinical Electronic Medical Records[J]. Interdisciplinary Sciences: Computational Life Sciences, 2024: 1-14.
[33] Wang Z, Qin Y, Chen W. Vision measurement of gear pitting based on DCGAN and U-Net[J]. Journal of Mechanical Science and Technology, 2021, 35: 2771-2779.
[34] Yang Z, Xu M, Wang S, et al. Detection of wind turbine blade abnormalities through a deep learning model integrating VAE and neural ODE[J]. Ocean Engineering, 2024, 302: 117689.
[35] Zhang Y, Zhou P, Ma E. Anomaly Detection of Industrial Smelting Furnace Incorporated With Accelerated Sampling Denoising Diffusion Probability Model and Conv-Transformer[J]. IEEE Transactions on Instrumentation and Measurement, 2024.