Research on Gearbox Fault Diagnosis Method based onParameter Optimized VMD and CNN
-
摘要: 针对齿轮箱的故障诊断的优化问题,提出了一种基于参数优化的变分模态分解(VMD)与卷积神经网络(CNN)相融合的故障诊断方法。该算法首先通过鲸鱼优化算法对VMD算法进行优化,之后通过正交实验法与粒子群优化算法进行了CNN模型中的重要参数进行优化,最后将分解后得到的固有模态分量输入CNN模型中进行训练学习。诊断完成后得到训练与检测结果,其中经过算法优化后CNN模型的训练与检测准确率可达98.7%与95.7%,优于未优化的准确率94.3%与91.8%。通过对结果的分析验证出该算法的可行性以及在诊断成功率方面的优越性,实现了故障特征信息的自适应性提取,并将故障类型进行分类,最终实现齿轮箱故障诊断的智能化。Abstract: Aiming at the optimization problem of gearbox fault diagnosis, a fault diagnosis method based on parameter optimization of variational mode decomposition (VMD) and convolutional neural network (CNN) is proposed. Firstly, the VMD algorithm is optimized by whale optimization algorithm. Then, the important parameters of CNN model are optimized by orthogonal experiment and particle swarm optimization algorithm. Next, the decomposed natural mode components are input into CNN model for training. After the diagnosis, the training and detection results are obtained. After the algorithm optimization, the training and detection accuracy of CNN model can reach 98.7% and 95.7% respectively, which is better than the non optimization accuracy of 94.3% and 91.8%. Through the analysis of the results, the feasibility of the proposed method and the superiority in the success rate of diagnosis are verified. The adaptive extraction of fault feature information is realized, and the fault types are classified. Finally, the intelligent fault diagnosis on gearboxis realized.
-
Key words:
- gearbox /
- fault diagnosis /
- variational mode decomposition /
- convolutional neural network
-
表 1 确定参数用数据样本
故障类型 转速/(r·min−1) 训练样本 测试样本 齿轮点蚀 1500 100 20 齿轮磨损 1500 100 20 齿轮变形 1500 100 20 表 2 正交实验结果
编号 A B C D 准确率/% 时间t 1 5 3 0.001 0.1 77.2 78.03 2 5 5 0.010 0.2 94.1 42.48 3 5 7 0.100 0.3 83.4 38.49 4 10 3 0.010 0.3 88.7 80.04 5 10 5 0.100 0.1 61.2 213.75 6 10 7 0.001 0.2 96.4 189.20 7 15 3 0.100 0.2 63.0 267.36 8 15 5 0.001 0.3 91.3 280.19 9 15 7 0.010 0.1 60.2 645.27 表 3 实验数据分析统计结果
参数 A B C D E1 254.7%(159.00) 228.9%(425.43) 264.9%(547.42) 198.6%(937.05) E2 246.3%(482.99) 246.6%(536.42) 243.0%(767.79) 253.5%(499.04) E3 214.5%(1192.87) 240.0%(872.96) 207.6%(519.60) 263.4%(398.72) e1 84.9%(53.00) 76.3%(141.81) 88.3%(182.47) 66.2%(312.35) e2 82.1%(161.00) 82.2%(178.81) 81.0%(255.93) 84.5%(166.35) e3 71.5%(397.62) 80.0%(290.99) 69.2%(173.20) 87.8%(132.91) Q 13.4%(344.62) 5.9%(149.18) 19.1%(82.73) 21.6%(179.44) 表 4 齿轮箱在齿轮故障中的CNN诊断成功率结果对比
优化结果 文献[17] 正交试验 正交试验+
MOPSO诊断成功率/% 95.4 96.1 98.9 运行时间/s 163.5 154.4 166.7 表 5 训练实验结果统计
编号 样本个数 故障类型 诊断
成功率/%诊断偏差
平均值/mmA B C A B C 1 100 Z 96 94 99 − − − 2 100 V 97 95 99 − − − 3 100 N 96 93 98 − − − 4 100 M 95 95 99 − − − 5 100 J 97 94 99 − − − 6 100 K 95 95 98 − − − 7 500 Z1 − − − 0.85 1.04 0.63 8 500 Z2 − − − 0.74 1.10 0.65 9 500 Z3 − − − 0.82 0.99 0.61 10 500 Z4 − − − 0.74 1.11 0.58 11 500 Z5 − − − 0.81 0.98 0.59 12 500 Z6 − − − 0.77 1.03 0.66 13 500 V1 − − − 0.72 0.96 0.58 14 500 V2 − − − 0.76 1.05 0.65 15 500 V3 − − − 0.80 0.97 0.61 16 500 V4 − − − 0.82 1.04 0.62 17 500 V5 − − − 0.78 1.12 0.59 18 500 V6 − − − 0.81 1.01 0.65 表 6 检测实验结果统计
编号 样本个数 故障类型 诊断
成功率/%诊断偏差
平均值/mmA B C A B C 19 100 X1 94 92 95 − − − 20 100 X2 93 91 96 − − − 21 100 X3 94 92 96 − − − 22 100 X4 93 93 96 − − − 23 100 X5 94 91 95 − − − 24 100 X6 95 92 96 − − − 25 500 ZX1 − − − 1.34 1.72 0.98 26 500 ZX2 − − − 1.29 1.69 1.10 27 500 VX3 − − − 1.37 1.74 1.06 28 500 VX4 − − − 1.41 1.66 1.02 -
[1] 高天阳, 肖守讷, 杨冰, 等. 修正准静态叠加法下的齿轮箱箱体寿命预测[J]. 机械科学与技术, 2019, 38(3): 480-486GAO T Y, XIAO S N, YANG B, et al. Life prediction of gearbox body by modified quasi-static superposition method[J]. Mechanical Science and Technology for Aerospace Engineering, 2019, 38(3): 480-486 (in Chinese) [2] HUANG N E, SHEN Z, LONG S R, et al. The empirical mode decomposition and the Hilbert spectrum for nonlinear and non-stationary time series analysis[J]. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 1998, 454(1971): 903-995 doi: 10.1098/rspa.1998.0193 [3] 周陈林, 董绍江, 李玲, 等. 滚动轴承多状态特征信息的改进型卷积神经网络故障诊断方法[J]. 振动工程学报, 2020, 33(4): 854-860ZHOU C L, DONG S J, LI L, et al. Method to improve convolutional neural network in rolling bearing fault diagnosis with multi-state feature information[J]. Journal of Vibration Engineering, 2020, 33(4): 854-860 (in Chinese) [4] 李思琦, 蒋志坚. 基于EEMD-CNN的滚动轴承故障诊断方法[J]. 机械强度, 2020, 42(5): 1033-1038LI S Q, JIANG Z J. Fault diagnosis method of rolling bearing based on EEMD-CNN[J]. Journal of Mechanical Strength, 2020, 42(5): 1033-1038 (in Chinese) [5] LI Y B, SI S B, LIU Z L, et al. Review of local mean decomposition and its application in fault diagnosis of rotating machinery[J]. Journal of Systems Engineering and Electronics, 2019, 30(4): 799-814 doi: 10.21629/JSEE.2019.04.17 [6] WANG H X, MIAO Y H, YANG H L, et al. Adaptive carrier fringe pattern enhancement for wavelet transform profilometry through modifying intrinsic time-scale decomposition[J]. Applied Optics, 2020, 59(20): 6191-6202 doi: 10.1364/AO.395603 [7] 鞠华, 沈长青, 黄伟国, 等. 基于支持向量回归的轴承故障定量诊断应用[J]. 振动、测试与诊断, 2014, 34(4): 767-771 doi: 10.3969/j.issn.1004-6801.2014.04.030JU H, SHEN C Q, HUANG W G, et al. Quantitative diagnosis of bearing fault based on support vector regression[J]. Journal of Vibration, Measurement & Diagnosis, 2014, 34(4): 767-771 (in Chinese) doi: 10.3969/j.issn.1004-6801.2014.04.030 [8] GAN M, WANG C, ZHU C A. Construction of hierarchical diagnosis network based on deep learning and its application in the fault pattern recognition of rolling element bearings[J]. Mechanical Systems and Signal Processing, 2016, 72-73: 92-104 doi: 10.1016/j.ymssp.2015.11.014 [9] 曾雪琼. 基于卷积神经网络的变速器故障分类识别研究[D]. 广州: 华南理工大学, 2016ZENG X Q. Classification and recognition of transmission fault based on convolutional neural network[D]. Guangzhou: South China University of Technology, 2016 (in Chinese) [10] HAN T, LIU C, YANG W G, et al. A novel adversarial learning framework in deep convolutional neural network for intelligent diagnosis of mechanical faults[J]. Knowledge-Based Systems, 2019, 165: 474-487 doi: 10.1016/j.knosys.2018.12.019 [11] DRAGOMIRETSKIY K, ZOSSO D. Variational mode decomposition[J]. IEEE Transactions on Signal Processing, 2014, 62(3): 531-544 doi: 10.1109/TSP.2013.2288675 [12] XIAO D M, DING J K, LI X J, et al. Gear fault diagnosis based on kurtosis criterion VMD and SOM neural network[J]. Applied Sciences, 2019, 9(24): 5424 doi: 10.3390/app9245424 [13] MIRJALILI S, LEWIS A. The whale optimization algorithm[J]. Advances in Engineering Software, 2016, 95: 51-67 doi: 10.1016/j.advengsoft.2016.01.008 [14] FAN Q, CHEN Z J, LI Z, et al. A new improved whale optimization algorithm with joint search mechanisms for high-dimensional global optimization problems[J]. Engineering with Computers, 2021, 37(3): 1851-1878 doi: 10.1007/s00366-019-00917-8 [15] 张朝林, 范玉刚. CEEMD与卷积神经网络特征提取的故障诊断方法研究[J]. 机械科学与技术, 2019, 38(2): 178-183ZHANG C L, FAN Y G. Fault diagnosis of a bearing using feature extraction method based on CEEMD algorithm and CNN[J]. Mechanical Science and Technology for Aerospace Engineering, 2019, 38(2): 178-183 (in Chinese) [16] INCE T, KIRANYAZ S, EREN L, et al. Real-time motor fault detection by 1-D convolutional neural networks[J]. IEEE Transactions on Industrial Electronics, 2016, 63(11): 7067-7075 doi: 10.1109/TIE.2016.2582729 [17] 吴春志, 江鹏程, 冯辅周, 等. 基于一维卷积神经网络的齿轮箱故障诊断[J]. 振动与冲击, 2018, 37(22): 51-56WU C Z, JIANG P C, FENG F Z, et al. Faults diagnosis method for gearboxes based on a 1-D convolutional neural network[J]. Journal of Vibration and Shock, 2018, 37(22): 51-56 (in Chinese) [18] 方开泰, 刘民千, 周永道. 试验设计与建模[M]. 北京: 高等教育出版社, 2011: 81-114FANG K T, LIU M Q, ZHOU Y D. Design and modeling of experiments[M]. Beijing: Higher Education Press, 2011: 81-114 (in Chinese) [19] COELLO C A C, PULIDO G T, LECHUGA M S. Handling multiple objectives with particle swarm optimization[J]. IEEE Transactions on Evolutionary Computation, 2004, 8(3): 256-279 doi: 10.1109/TEVC.2004.826067