A Weighted GWKNN-SVM Method for Fault Diagnosis of Water Pumps
-
摘要: 针对实际运行环境下的工业水泵具有工况数据量大、运行时间长、特征类型多等特点,提出一种基于特征加权的高斯加权K近邻-支持向量机(GWKNN-SVM)的组合故障诊断分类算法。首先通过对某化工厂三台水泵5个月份的运行采集数据进行特征提取和清洗,然后分别使用高斯加权K近邻算法(GWKNN)-支持向量机算法(SVM)对数据进行快速粗分类和边界数据细分类,以提高水泵故障分类精度和识别效率。最后通过仿真实例比较了相同条件下GWKNN-SVM算法和其他分类算法的故障分类效果。试验结果表明,该组合分类方法能够有效提高水泵工况的故障分类精度,从而实现工业环境下的水泵健康监测。Abstract: Aiming at the characteristics of industrial water pumps in actual operating environment, such as large amount of working condition data, long running time and multiple feature types, a combined fault diagnosis classification algorithm based on the weighted Gaussian weighted K nearest neighbor and the support vector machine (GWKNN-SVM) is proposed. First, feature extraction and cleaning of given industrial data which is collected from three pumps of a chemical plant from 5 months are performed, and then the Gaussian weighted K nearest neighbor algorithm (GWKNN) is used to classify the data quickly and coarsely. Moreover, the support vector machine algorithm (SVM) is used to classify the boundary data which is chosen by GWKNN for fine classification to improve the classification accuracy and identification efficiency of pump faults. Finally, a simulation example is used to compare the fault classification effects using the GWKNN-SVM algorithm and other classification algorithms under the same conditions. Experimental results show that the combined classification method proposed can improve the fault classification accuracy of the water pump effectively, and then achieves the object of the pump health monitoring in the industrial environment.
-
Key words:
- K-nearest neighbor algorithm /
- support vector machine /
- feature weighting /
- water pump /
- fault diagnosis
-
表 1 水泵工况类型数据表
水泵型号 电机振动/(mm·s−1) 水泵振动/(mm·s−1) 电机转速/(r·min−1) 电机温度/℃ 水泵工况
GSSB10.38181752 1.407695293 1003.125 18.8120842 B 0.208443925 0.364216268 1246.875 17.054245 C 0.616366863 0.47196728 2400 47.09925842 D 0.259716809 0.090452619 3000 21.32677269 E GSSB2 0.232338 1.388318 1275 24.4976 B 0.244651 1.414074 1350 22.11109 C LSSB1 0.742220283 0.513350725 2400 47.17860794 D 0.259716809 0.090452619 3000 21.32677269 E 表 2 权值及SVM训练参数
数据集 权值 C g 核函数 模型 GSSB1-4 0.2 100 5 rbf ovo GSSB1-6 0.4 200 5 rbf ovo GSSB1-8 0.7 300 5 rbf ovo GSSB2 0.3 150 10 rbf ovo LSSB1 0.6 250 20 rbf ovo 表 3 水泵工况故障表现及原因
工况代号 故障表现 主要原因 A 无法启动、
自动断电异物堵塞、线路短路
泵轴形变、泵内锈蚀B 泵体异常抖动、
流量不足电机滚珠轴承损坏、
密封环或叶轮磨损过大C 流量不足 密封环或叶轮磨损过大 D 电机过热 运行时间过长、缺少
润滑油、轴承破裂E 正常运行 泵体正常 表 4 不同分类器准确率结果对比表
% 分类器 GSSB1-8 GSSB2 LSSB1 NB 80 91 87 ANN 86 92.16 90 SVM 91.97 90.87 95.26 KNN 91.65 87.52 95.55 GA-SVM 92.01 92.40 95.34 GWKNN-SVM 93.03 92.60 95.75 -
[1] 马双忱, 马岚, 刘畅, 等. 电厂循环冷却水处理技术研究与应用进展[J]. 化学工业与工程, 2019, 36(1): 38-47MA S C, MA L, LIU C, et al. Progress in research and application of circulating cooling water treatment technologies in power plants[J]. Chemical Industry and Engineering, 2019, 36(1): 38-47 (in Chinese) [2] ZHANG H S, ZHAO H B, LI Z L, et al. Optimization potentials for the waste heat recovery of a gas-steam combined cycle power plant based on absorption heat pump[J]. Journal of Thermal Science, 2019, 28(2): 283-293 [3] RAPUR J S, TIWARI R. On-line time domain vibration and current signals based multi-fault diagnosis of centrifugal pumps using support vector machines[J]. Journal of Nondestructive Evaluation, 2019, 38(1): 6 [4] NIKRAVESH S M Y, REZAIE H, KILPATRIK M, et al. Intelligent fault diagnosis of bearings based on energy levels in frequency bands using wavelet and support vector machines (SVM)[J]. Journal of Manufacturing and Materials Processing, 2019, 3(1): 11 [5] ZHANG C, LI C S, PENG T, et al. Modeling and synchronous optimization of pump turbine governing system using sparse robust least squares support vector machine and hybrid backtracking search algorithm[J]. Energies, 2018, 11(11): 3108 [6] 孙斌, 薛广鑫. 基于等距特征映射和支持矢量机的转子故障诊断方法[J]. 机械工程学报, 2012, 48(9): 129-135SUN B, XUE G X. Method of rotor fault diagnosis based on isometric feature mapping and support vector machine[J]. Journal of Mechanical Engineering, 2012, 48(9): 129-135 (in Chinese) [7] 宋涛, 舒涛, 刘进, 等. 基于改进最小二乘支持向量机的柱塞泵故障诊断模型[J]. 探测与控制学报, 2015, 37(4): 91-96SONG T, SHU T, LIU J, et al. A fault diagnosis for plunger pump based on improved LSSVM[J]. Journal of Detection & Control, 2015, 37(4): 91-96 (in Chinese) [8] CAI B P, LIU Y H, FAN Q, et al. Multi-source information fusion based fault diagnosis of ground-source heat pump using Bayesian network[J]. Applied Energy, 2014, 114: 1-9 [9] LIU Z K, LIU Y H, SHAN H K, et al. A fault diagnosis methodology for gear pump based on EEMD and Bayesian network[J]. PLoS One, 2015, 10(5): e0125703 [10] 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) [11] 焦晓璇, 景博, 李娟, 等. 基于ADEGWO-SVM的机载燃油泵寿命预测研究[J]. 仪器仪表学报, 2018, 39(8): 43-52JIAO X X, JING B, LI J, et al. Research on remaining useful life prediction of fuel pump based on adaptive differential evaluation grey wolf optimizer-support vector machine[J]. Chinese Journal of Scientific Instrument, 2018, 39(8): 43-52 (in Chinese) [12] LI K, WANG L Y, WU J J, et al. Using GA-SVM for defect inspection of flip chips based on vibration signals[J]. Microelectronics Reliability, 2018, 81: 159-166 [13] GU Y K, ZHOU X Q, YU D P, et al. Fault diagnosis method of rolling bearing using principal component analysis and support vector machine[J]. Journal of Mechanical Science and Technology, 2018, 32(11): 5079-5088 [14] ZHANG Y S, GAO Q W, LU Y X, et al. A novel intelligent method for bearing fault diagnosis based on hermitian scale-energy spectrum[J]. IEEE Sensors Journal, 2018, 18(16): 6743-6755 doi: 10.1109/JSEN.2018.2851100 [15] CHEN F F, TANG B P, CHEN R X. A novel fault diagnosis model for gearbox based on wavelet support vector machine with immune genetic algorithm[J]. Measurement, 2013, 46(1): 220-232 [16] SHANG H K, XU J Y, ZHENG Z T, et al. A novel fault diagnosis method for power transformer based on dissolved gas analysis using hypersphere multiclass support vector machine and improved D-S evidence theory[J]. Energies, 2019, 12(20): 4017 doi: 10.3390/en12204017 [17] SAMIGULINA G A, SAMIGULINA Z I. Modified immune network algorithm based on the random forest approach for the complex objects control[J]. Artificial Intelligence Review, 2019, 52(4): 2457-2473 doi: 10.1007/s10462-018-9621-7 [18] 姚德臣, 杨建伟, 程晓卿, 等. 基于多尺度本征模态排列熵和SA-SVM的轴承故障诊断研究[J]. 机械工程学报, 2018, 54(9): 168-176YAO D C, YANG J W, CHENG X Q, et al. Railway rolling bearing fault diagnosis based on muti-scale IMF permutation entropy and SA-SVM classifier[J]. Journal of Mechanical Engineering, 2018, 54(9): 168-176 (in Chinese) [19] YEH J P, CHIANG C M. Reducing the solution of support vector machines using simulated annealing algorithm[C]//2017 International Conference on Control, Artificial Intelligence, Robotics & Optimization (ICCAIRO). Prague: IEEE, 2017: 105-108 [20] LIU Q B, SHI W K, CHEN Z Y. Fatigue life prediction for vibration isolation rubber based on parameter-optimized support vector machine model[J]. Fatigue & Fracture of Engineering Materials & Structures, 2019, 42(3): 710-718 [21] AZADEH A, SABERI M, KAZEM A, et al. A flexible algorithm for fault diagnosis in a centrifugal pump with corrupted data and noise based on ANN and support vector machine with hyper-parameters optimization[J]. Applied Soft Computing, 2013, 13(3): 1478-1485 [22] DENOEUX T. A k-nearest neighbor classification rule based on dempster-shafer theory[J]. IEEE Transactions on Systems, Man, and Cybernetics, 1995, 25(5): 804-813 [23] YIGIT H. ABC-based distance-weighted KNN algorithm[J]. Journal of Experimental & Theoretical Artificial Intelligence, 2015, 27(2): 189-198