基于纹理分析的柴油发动机故障诊断方法

doi:10.12382/bgxb.2022.1021

摘要/Abstract

摘要：

柴油发动机故障特征提取是对其故障识别过程中的关键步骤,直接关系到识别的准确性和时效性。将纹理分析理论应用于柴油发动机故障特征提取,提出基于改进层次分解(Modified Hierarchical Decomposition, MHD)和灰度图像处理的故障特征提取方法。使用MHD将单个一维的振动信号样本分解为多个一维子信号并分别转换成灰度图像;加速分割测试获得特征(Features from Accelerated Segment Test, FAST)算法被用于检测灰度图像的特征点;通过Gabor滤波器组的实部对图像进行卷积,利用特征点的响应计算直方图作为特征向量。为检验通过该方法提取的故障特征对柴油发动机不同故障类型的识别能力,引入非支配排序遗传算法Ⅱ(Non-dominated Sorting Genetic Algorithm-Ⅱ, NSGA-Ⅱ)和支持向量机(Support Vector Machine, SVM)进行故障状态识别。通过实验台开展预置故障实验,将提出的方法与传统方法对比。实验结果表明:该方法的故障诊断准确率最高,为开展柴油发动机故障诊断提供一种新的思路。

关键词: 柴油发动机, 纹理分析, 改进层次分解, 非支配排序遗传算法Ⅱ, 支持向量机

Abstract:

Diesel engine fault feature extraction is a key step in the process of fault identification, which is directly related to the accuracy and timeliness of identification. The texture analysis theory is applied to diesel engine fault feature extraction, and a fault feature extraction method based on modified hierarchical decomposition (MHD) and grayscale image processing is proposed. A single one-dimensional vibration signal sample is decomposed into multiple one-dimensional sub-signals and converted into a grayscale image separately using MHD. The features from accelerated segment test (FAST) algorithm is used to detect the feature points of grayscale image; the image is convolved by the real part of Gabor filter bank, and the histograms are computed as feature vectors using the responses of the feature points. In order to test the ability of the fault features extracted by the proposed method to recognize the different fault types of diesel engine, non-dominated sorting genetic algorithm-Ⅱ (NSGA-Ⅱ) and support vector machine (SVM) are introduced for fault status recognition. Preset fault experiments are carried out through a experimental bench to compare the proposed method with the traditional method. The experimental results show that the proposed method has the highest fault diagnosis accuracy and provides a new idea for diesel engine fault diagnosis.

Key words: diesel engine, texture analysis, modified hierarchical decomposition, non-dominated sorting genetic algorithm-Ⅱ, support vector machine

中图分类号:

TH113.1

刘子昌, 李思雨, 裴模超, 刘洁, 孟硕, 吴巍屹. 基于纹理分析的柴油发动机故障诊断方法[J]. 兵工学报, 2024, 45(2): 684-694.

LIU Zichang, LI Siyu, PEI Mochao, LIU Jie, MENG Shuo, WU Weiyi. Fault Diagnosis Method for Diesel Engine Based on Texture Analysis[J]. Acta Armamentarii, 2024, 45(2): 684-694.

图/表 16

图1 MHD信号示例

Fig.1 An illustration for decomposing signal of MHD

图2 FAST算法特征点检测

Fig.2 Feature point detection using FAST algorithm

图3 一维信号转换成二维矩阵

Fig.3 Conversion of one-dimensional signal into two-dimensional matrix

图4 本文提出的特征提取方法示例

Fig.4 Example of proposed feature extraction method

图5 NSGA-Ⅱ-SVM对特征选择和状态识别的联合优化流程

Fig.5 Flowchart of joint optimization for feature selection and state identification using NSGA-Ⅱ-SVM

图6 柴油机状态监测实验台

Fig.6 Diesel engine condition monitoring test bench

表1 柴油发动机预置故障模式

Table 1 Diesel engine preset fault mode

序号	故障模式	序号	故障模式
1	正常状态	4	空气滤清器堵塞
2	一缸失火	5	喷油泵供油不足
3	二缸失火	6	一、二缸失火

图7 6种故障状态时域波形

Fig.7 Time-domain waveforms for 6 fault statuses

图8 各子信号的时域波形(上)和图像(下)

Fig.8 Time-domain waveform(above) and image representation(below) of each sub-signal

图9 t取不同值时各个故障类型的特征点数

Fig.9 The number of feature points of the type of failure at different t

图10 从两个角度展示的分类精度与ksize、σ和t三个参数之间的可视化关系

Fig.10 Visualizable relationships among classification accuracy and the parameters ksize, σ and t shown from two angles

图11 ksize、σ和t取不同值时的分类精度

Fig.11 Classification accuracy with different values of ksize, σ and t

图12 实例2中4种方法的识别准确率

Fig.12 Identification accuracies of 4 methods in Instance 2

表2 实例2中4种方法的识别准确率

Table 2 Identification accuracy statistics of 4 methods in Instance 2 %

方法	最大值	最小值	平均值
1	85.59	84.26	85.06
2	88.07	86.03	86.79
3	97.34	96.07	97.05
4	95.28	93.71	94.68

图13 实例3中4种方法的比较结果

Fig.13 Comparison results of 4 methods in Instance 3

表3 实例3中4种方法的故障识别结果

Table 3 Fault status identification results of 4 methods in Instance 3 %

方法	最大值	最小值	平均值
本文方法	97.76(11)	96.17(8)	97.18
LBP-NSGA-Ⅱ-SVM	95.61(19)	93.22(15)	94.27
1D-LTP-NSGA-Ⅱ-SVM(上模式)	97.08(16)	95.16(12)	96.34
1D-LTP-NSGA-Ⅱ-SVM(下模式)	96.63(14)	94.73(12)	95.51

参考文献 23

[1]	杨攀涛, 崔涛, 赵彦凯. 电动增压器对柴油发动机低速稳态性能的影响[J]. 兵工学报, 2021, 42(9): 1829-1837. doi: 10.3969/j.issn.1000-1093.2021.09.003
	YANG P T, CUI T, ZHAO Y K. Effect of electric supercharger on low-speed steady-state performance of diesel engine[J]. Acta Armamentarii, 2021, 42(9): 1829-1837. (in Chinese)
[2]	仝蕊, 康建设, 孙健, 等. 基于局部特征尺度分解与复合谱分析的齿轮性能退化特征提取[J]. 兵工学报, 2019, 40(5): 1093-1102. doi: 10.3969/j.issn.1000-1093.2019.05.023
	TONG R, KANG J S, SUN J, et al. Gear performance degrada-tion feature extraction based on local characteristic-scale de-composition and modified composite spectrum analysis[J]. Acta Armamentarii, 2019, 40(5): 1093-1102. (in Chinese) doi: 10.3969/j.issn.1000-1093.2019.05.023
[3]	杨大为, 赵永东, 冯辅周, 等. 基于参数优化变分模态分解和多尺度熵偏均值的行星变速箱故障特征提取[J]. 兵工学报, 2018, 39(9): 1683-1691. doi: 10.3969/j.issn.1000-1093.2018.09.003
	YANG D W, ZHAO Y D, FENG F Z, et al. Planetary gearbox fault feature extraction based on parameter optimized varia-tional mode decomposition and partial mean of multi-scale en-tropy[J]. Acta Armamentarii, 2018, 39(9): 1683-1691. (in Chinese)
[4]	李良钰, 苏铁熊, 马富康, 等. 基于集合经验模态分解-支持向量机的高压共轨系统故障诊断方法[J]. 兵工学报, 2022, 43(5): 992-1001. doi: 10.12382/bgxb.2021.0155
	LI L Y, SU T X, MA F K, et al. Fault diagnosis method of high-pressure common rail system based on EEMD-SVM[J]. Acta Armamentarii, 2022, 43(5): 992-1001. (in Chinese) doi: 10.12382/bgxb.2021.0155
[5]	饶响, 盛晨兴, 郭智威. 不同表面纹理结构对柴油机缸套-活塞环摩擦磨损性能的影响研究[J]. 兵工学报, 2018, 39(2): 356-363. doi: 10.3969/j.issn.1000-1093.2018.02.019
	RAO X, SHENG C X, GUO Z W. Study on the influence of different surface texture structure on friction and wear per-formance of cylinder liner-piston ring of diesel engine[J]. Acta Armamentarii, 2018, 39(2): 356-363. (in Chinese)
[6]	KAPLAN K, KAYA Y, KUNCAN M, et al. An improved feature ex-traction method using texture analysis with LBP for bearing fault diagnosis[J]. Applied Soft Computing, 2020, 87: 106019. doi: 10.1016/j.asoc.2019.106019 URL
[7]	ZHENG H, CHENG G, LI Y, et al. A new fault diagnosis method for planetary gear based on image feature extraction and bag-of-words model[J]. Measurement, 2019, 145: 1-13. doi: 10.1016/j.measurement.2019.05.046 URL
[8]	KUNCAN M, KAPLAN K, MⅰNAZ M R, et al. A novel feature ex-traction method for bearing fault classification with one dimen-sional ternary patterns[J]. ISA Transactions, 2020, 100: 346-357. doi: 10.1016/j.isatra.2019.11.006 URL
[9]	RUIZ M, MUJICA L E, ALFEREZ S, et al. Wind turbine fault de-tection and classification by means of image texture analysis[J]. Mechanical Systems and Signal Processing, 2018, 107: 149-167. doi: 10.1016/j.ymssp.2017.12.035 URL
[10]	彭泊涵, 欧阳杨, 张良. 几何与Gabor纹理特征的机载LiDAR点云建筑物提取[J]. 测绘科学, 2022, 47(6):119-126.
	PENG B H, OUYANG Y, ZHANG L. Airborne LiDAR point cloud building extraction with geometric and Gabor texture features[J]. Science of Surveying and Mapping, 2022, 47(6):119-126. (in Chinese)
[11]	LI Y B, LI G Y, YANG Y T, et al. A fault diagnosis scheme for planetary gearboxes using adaptive multi-scale morphology filter and modified hierarchical permutation entropy[J]. Me-chanical Systems and Signal Processing, 2018, 105: 319-337.
[12]	YANG C, JIA M P. Health condition identification for rolling bear-ing based on hierarchical multiscale symbolic dynamic entropy and least squares support tensor machine-based binary tree[J]. Structural Health Monitoring, 2021, 20(1): 151-172. doi: 10.1177/1475921720923973 URL
[13]	卢胜男, 李小和. 基于对称FAST特征的车辆目标检测方法[J]. 微电子学与计算机, 2020, 37(2):37-42.
	LU S N, LI X H. Vehicle target detection method based on symmetric FAST feature[J]. Microelectronics and Com-puter, 2020, 37(2):37-42. (in Chinese)
[14]	VIJAYAN T, SANGEETHA M, KUMARAVEL A, et al. Gabor filter and machine learning based diabetic retinopathy analysis and detection[J]. Microprocessors and Microsystems, 2020: 103353.
[15]	WU J H, WEI P J, YUAN X F, et al. A new Gabor filter-based method for automatic recognition of hatched residential ar-eas[J]. IEEE Access, 2019: 40649-40662.
[16]	ROSTEN E, PORTER R, DRUMMOND T. Faster and better: a machine learning approach to corner detection[J]. IEEE Trans-actions on Software Engineering, 2010, 32(1): 105-119.
[17]	BIANCONI F, FERNÁNDEZ A. Evaluation of the effects of Gabor filter parameters on texture classification[J]. Pattern recogni-tion, 2007, 40(12): 3325-3335.
[18]	ZHANG Y K, LI W J, ZHANG L P, et al. Adaptive learning Gabor filter for finger-vein recognition[J]. IEEE Access, 2019, 7: 159821-159830. doi: 10.1109/ACCESS.2019.2950698
[19]	BAI Y L, YANG J W, WANG J H, et al. Image representation of vi-bration signals and its application in intelligent compound fault diagnosis in railway vehicle wheelset-axlebox assemblies[J]. Mechanical Systems and Signal Processing, 2021, 152: 107421. doi: 10.1016/j.ymssp.2020.107421 URL
[20]	ZHANG W, PENG G L, LI C H. Bearings fault diagnosis based on convolutional neural networks with 2-D representation of vi-bration signals as input[C]//Proceedings of the 2016 the 3rd In-ternational Conference on Mechatronics and Mechanical Engi-neering. Shanghai, China: EDP Sciences, 2017, 95: 13001.
[21]	WANG S P, ZHAO D M, YUAN J Z, et al. Application of NSGA-Ⅱ algorithm for fault diagnosis in power system[J]. Electric Power Systems Research, 2019, 175:105893. doi: 10.1016/j.epsr.2019.105893 URL
[22]	JIA W, HU R X, LEI Y K, et al. Histogram of oriented lines for palmprint recognition[J]. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 2014, 44(3): 385-395. doi: 10.1109/TSMC.6221021 URL
[23]	KAPLAN K, KAYA Y, KUNCAN M, et al. An improved feature ex-traction method using texture analysis with LBP for bearing fault diagnosis[J]. Applied Soft Computing, 2020, 87:106019. doi: 10.1016/j.asoc.2019.106019 URL