为进一步提高风电齿轮箱故障诊断的准确率,提出一种格拉姆角差场下基于改进卷积神经网络（CNN）和极限梯度提升（XGBoost）融合的智能诊断模型。首先,通过格拉姆角差场变换,将复杂环境下齿轮箱振动信号的一维数据转换为二维图像,完整保留信号内在结构和时频特征;其次,提出一种改进CNN的齿轮箱振动图像多维特征提取方法,在传统CNN的卷积层引入双注意力模块提升模型的感知能力,加大全局信息的提取程度,采用改进βc-ACONC激活函数代替ReLU激活函数,对神经元进行选择性激活,提升网络整体特征的表达能力;然后,将提取的综合特征输入改进麻雀搜索算法（ISSA）超参优化的XGBoost网络,构建一种格拉姆角差场下基于改进CNN-XGBoost融合的风电齿轮箱故障诊断模型。最后,采用实验室风电机组齿轮箱数据集对模型性能进行验证,试验结果表明模型诊断精度高达99%以上,具有良好的故障识别能力。
Abstract
To further enhance the accuracy of fault diagnosis for wind turbine gearboxes, an intelligent diagnostic model based on the improved convolutional neural network （CNN） and extreme gradient boosting （XGBoost） fusion under the Gramian angular difference field （GADF） transformation is proposed. Firstly, the one-dimensional vibration signal data of the gearbox under complex environments are transformed into two-dimensional images through the GADF transformation, which retains the intrinsic structure and time-frequency characteristics of the signals. Secondly, an improved CNN method for multi-dimensional feature extraction of gearbox vibration images is proposed, where dual attention modules are introduced in the convolutional layers of the traditional CNN to enhance the model’s perceptual capabilities and increase the extraction of global information. The improvedβc-ACONC activation function which replaces the ReLU activation function is employed to selectively activate neurons, thereby enhancing the network's overall feature expression capabilities. Then, the integrated features are input into the XGBoost network optimized by ISSA hyperparameters, constructing a wind turbine gearbox fault diagnostic model based on the improved CNN-XGBoost fusion under the GADF. Finally, the performance of the model is verified using a laboratory wind turbine gearbox dataset, and the experimental results demonstrate that the model has a diagnostic accuracy of over 99%, exhibiting excellent fault recognition capabilities.
关键词
风力发电机 /
卷积神经网络 /
故障诊断 /
格拉姆角差场 /
极限梯度提升 /
齿轮箱
Key words
wind turbines /
convolutional neural networks /
fault diagnosis /
Gramian angular difference field /
XGBoost /
gearbox
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
参考文献
[1] 代东昌, 周建星, 张建杰. 齿轮传动系统碰撞振动特性研究[J]. 机械设计与制造, 2019(1): 8-12.
Dai D C, Zhou J X, Zhang J J.Vibro-impact characteristics of a gear train[J]. Machinery Design & Manufacture, 2019(1): 8-12.
[2] 赵东旭, 张彬彬, 马子魁. 风电齿轮箱行星齿轮滑动轴承的润滑性能[J]. 轴承, 2024(9): 28-36.
Zhao D X, Zhang B B, Ma Z K.Lubrication performance of sliding bearings for planetary gears in wind turbine gearboxes[J]. Bearing, 2024(9): 28-36.
[3] 咸博. 基于小波分析的风电机组齿轮箱振动信号识别方法设计[J]. 机械管理开发, 2024, 39(5): 146-147, 163.
Xian B.Design of wavelet analysis based vibration signal identification method for wind turbine gearboxes[J]. Mechanical Management and Development, 2024, 39(5): 146-147, 163.
[4] 候双珊, 郑近德, 潘海洋, 等. 基于复合多尺度交叉模糊熵的行星齿轮箱故障诊断[J]. 振动与冲击, 2023, 42(20): 130-135, 171.
Hou S S, Zheng J D, Pan H Y, et al.Planetary gearbox fault diagnosis based on composite multi-scale cross fuzzy entropy[J]. Journal of Vibration and Shock, 2023, 42(20): 130-135, 171.
[5] 花志锋, 赵荣珍. 风电齿轮箱行星轮系的齿面摩擦与齿根裂纹耦合效应计算方法[J]. 太阳能学报, 2022, 43(9): 287-293.
Hua Z F, Zhao R Z.Calculation method of coupling effect of tooth surface friction and tooth root cracking in planetary wheel systems for wind turbine gearboxes[J]. Acta Energiae Solaris Sinica, 2022, 43(9): 287-293.
[6] 毕玉, 朱才朝, 谭建军, 等. 齿根裂纹对风电齿轮箱高速级动态特性影响分析及其故障模式识别[J]. 太阳能学报, 2022, 43(7): 284-292.
Bi Y, Zhu C C, Tan J J, et al.Influence of tooth root crack on dynamic characteristics of wind turbine gearbox high-speed stage and failure mode identification[J]. Acta Energiae Solaris Sinica, 2022, 43(7): 284-292.
[7] Jiang S, Li W, Xin G F, et al.Analysis of torsional vibration characteristics and time delay feedback control of semi-direct drive cutting transmission system in shearer[J]. Chaos, Solitons & Fractals, 2020, 132: 109607.
[8] 张文斌, 江洁, 俞利宾, 等. 互补集合经验模式分解与奇异值能量谱在风电齿轮故障识别中的应用[J]. 太阳能学报, 2020, 41(2): 137-143.
Zhang W B, Jiang J, Yu L B, et al.Application of complementary ensemble empirical mode decomposition and singular value energy spectrum in wind power gear fault identification[J]. Acta Energiae Solaris Sinica, 2020, 41(2): 137-143.
[9] 林近山, 窦春红, 寇兴磊. 基于时变滤波经验模式分解的齿轮箱故障诊断[J]. 机械传动, 2018, 42(1): 98-101, 105.
Lin J S, Dou C H, Kou X L.Fault diagnosis of gearboxes based on time varying filter for empirical mode decomposition[J]. Journal of Mechanical Transmission, 2018, 42(1): 98-101, 105.
[10] 张德祥, 汪萍, 吴小培, 等. 基于经验模式分解和Teager能量谱的齿轮箱故障诊断[J]. 振动与冲击, 2010, 29(7): 109-111, 138.
Zhang D X, Wang P, Wu X P, et al.Gearbox fault diagnosis based on empirical mode decomposition and Teager energy spectrum[J]. Journal of Vibration and Shock, 2010, 29(7): 109-111, 138.
[11] 许冬, 关大志, 田武成, 等. 风力发电机组齿轮箱振动监测与故障诊断方法研究[J]. 中国设备工程, 2024(23): 198-200.
Xu D, Guan D Z, Tian W C, et al.Research on vibration monitoring and fault diagnosis methods for wind turbine gearboxes[J]. China Plant Engineering, 2024(23): 198-200.
[12] 陈志强, 陈旭东, José Valente de Olivira, 等. 深度学习在设备故障预测与健康管理中的应用[J]. 仪器仪表学报, 2019, 40(9): 206-226.
Chen Z Q, Chen X D, José Valente de Olivira, et al. Application of deep learning in equipment prognostics and health management[J]. Chinese Journal of Scientific Instrument, 2019, 40(9): 206-226.
[13] 范亚飞, 郝如江, 杨青松, 等. 基于VMD和AVOA-SCN的齿轮箱故障诊断[J]. 制造技术与机床, 2023(12): 19-25.
Fan Y F, Hao R J, Yang Q S, et al.Gearbox fault diagnosis based on VMD and AVOA-SCN[J]. Manufacturing Technology & Machine Tool, 2023(12): 19-25.
[14] 徐洁. 基于CEEMDAN-CNN-BiLSTM的行星齿轮箱故障诊断分析[J]. 电子技术, 2023, 52(9): 414-415.
Xu J.Analysis of planetary gearbox fault diagnosis based on CEEMDAN-CNN-BiLSTM[J]. Electronic Technology, 2023, 52(9): 414-415.
[15] Jiang G Q, He H B, Yan J, et al.Multiscale convolutional neural networks for fault diagnosis of wind turbine gearbox[J]. IEEE Transactions on Industrial Electronics, 2019, 66(4): 3196-3207.
[16] Gai J B, Shen J X, Wang H, et al.A parameter-optimized DBN using GOA and its application in fault diagnosis of gearbox[J]. Shock and Vibration, 2020, 2020: 4294095.
[17] Bao W J, Liu Z, Liu S Y, et al.Application of high-order multisynchrosqueezing transform in fault diagnosis[J]. IEEE Transactions on Instrumentation and Measurement, 2024, 73: 3529410.
[18] 赵妍, 唐文石, 聂永辉, 等. 基于格拉姆角差场和卷积神经网络的宽频振荡分类方法[J]. 电网技术, 2022, 46(11): 4364-4372.
Zhao Y, Tang W S, Nie Y H, et al.Broadband oscillation classification method based on GADF-AlexNet[J]. Power System Technology, 2022, 46(11): 4364-4372.
[19] 张嘉雯, 蔡彬彬, 林崧. 基于粒子群优化算法的量子卷积神经网络[J]. 量子电子学报, 2025, 42(1): 123-135.
Zhang J W, Cai B B, Lin S.Quantum convolutional neural network based on particle swarm optimization algorithm[J]. Chinese Journal of Quantum Electronics, 2025, 42(1): 123-135.
[20] Woo S, Park J, Lee J Y, et al.CBAM: convolutional block attention module[C]//Proceedings of the European Conference on Computer Vision(ECCV), 2018: 3-19.
[21] Ma N, Zhang X, Liu M, et al.Activate or not: learning customized activation[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Nashville, TN, USA, 2021: 8028-8038.
[22] Xue J, Shen B.A novel swarm intelligence optimization approach: sparrow search algorithm[J]. Systems Science & Control Engineering, 2020, 8(1): 22-34.
[23] 陈诺, 吕干云, 叶加星. 基于SVM级联决策树的复合电能质量扰动识别[J]. 电气工程学报, 2023(2): 149-156.
Chen N, Lyu G Y, Ye J X.Recognition of complex PQ disturbances based on SVM cascaded decision tree[J]. Journal of Electrical Engineering, 2023(2): 149-156.
基金
国家重点研发计划（2022YFB3105101）
PDF(2910 KB)
