Page 116 - 《摩擦学学报》2021年第1期

第 1 期 杨智宏, 等: 基于Mask R-CNN网络的磨损颗粒智能识别与应用 113

0 39 1 1 0 0 0 0 0 0 depending on debris micrology shape analysis[J]. Journal of Basic
35
0 27 0 0 1 0 0 0 0
1 Science and Engineering, 2000, 8(4): 431–437 (in Chinese) [李艳军,
0 0 32 0 0 0 0 0 0 30
2 左洪福, 吴振锋. 基于磨粒显微形态分析的发动机磨损状态监测
0 0 0 25 0 0 0 0 0 25
5 True label 4 0 0 2 0 2 0 1 0 30 25 1 0 0 1 0 1 20 431–437]. doi: 10.3969/j.issn.1005-0930.2000.04.013.
3 与 故 障 诊 断 技 术 [J]. 应 用 基 础 与 工 程 科 学 学 报 , 2000, 8(4):
2
0
0 0 0 1 0 1 37 0 0 15 [ 7 ] Zhou Xincong, Xiao Hanliang, Yan Xinping, et al. A new
6 10 comprehensive feature parameter for wear particle image analysis[J].
0 0 0 0 0 0 0 35 0
7 Tribology, 2002, 22(2): 138–141 (in Chinese) [周新聪, 萧汉梁, 严
5
0 0 0 0 0 0 0 1 32
8 新平, 等. 一种新的磨粒图像特征参数[J]. 摩擦学学报, 2002,
0
0 1 2 3 4 5 6 7 8 22(2): 138–141]. doi: 10.3321/j.issn:1004-0595.2002.02.014.
Predicted label
[ 8 ] Fan Hongwei, Ding Xiao, Gao Shuoqi, et al. Abrasive particle

Fig. 9 Confusion matrix for the prediction result
feature extraction in ferrography based on binary correction
of wear index
图 9 磨损指数预测的混淆矩阵 ofinverse grayscale image[J]. Lubrication Engineering, 2019, 44(6):
66–71 (in Chinese) [樊红卫, 丁骁, 高烁琪, 等. 基于反相灰度图二
现磨粒的像素级识别，通过对识别结果处理可数字化 值化修正的铁谱图像磨粒特征提取[J]. 润滑与密封, 2019, 44(6):
表征出每个颗粒的具体面积和尺寸，解决了磨粒分析 66–71]. doi: 10.3969/j.issn.0254－0150.2019.06.010.
[ 9 ] Kong Xiangxing, Shao Tao. Wear debris material recognition based
中难定量的问题，对磨粒分析的智能化、自动化有实
on color feature extraction[J]. Lubrication Engineering, 2020, 45(5):
际的指导意义.
79–85 (in Chinese) [孔祥兴, 邵涛. 基于颜色特征提取的磨粒材质
参 考 文 献 识别研究[J]. 润滑与密封, 2020, 45(5): 79–85]. doi: 10.3969/j.issn.
0254-0150.2020.05.013.
[ 1 ] Feng Wei, Li Qiuqiu, He Shizhong. Method and application of
[10] Peng Yeping, Wu Tonghai, Cao Guangzhong, et al. A hybrid search-
particle classification based on analytical ferrography[J]. Lubrication
tree discriminant technique for multivariate wear debris
Engineering, 2015(12): 125–130 (in Chinese) [冯伟, 李秋秋, 贺石
classification[J]. Wear, 2017, 392/393: 152–158. doi: 10.1016/j.wear.
中. 基于铁谱分析的颗粒分类识别方法与应用[J]. 润滑与密封,
2017.09.022.
2015(12): 125–130]. doi: 10.3969/j.issn.0254-0150.2015.12.024.
[11] Peerawatt Nunthavarawong. Comparative study on wear particle
[ 2 ] Liang Hua, Yang Mingzhong. Wear particle identification expert
colour classifications using various machine learning algorithms[J].
system based on neural network [C] / / China Society of Mechanical
Applied Mechanics and Materials, 2014, 619: 347–351. doi:
Engineering 94 National Ferrography Technology Conference,
10.4028/www.scientific.net/AMM.619.347.
1994（in Chinese） [梁华, 杨明忠. 基于神经网络的磨粒识别专家
[12] LeCun Y, Bengio Y, Hinton G. Deep learning[J]. Nature, 2015,
系统[C]//中国机械工程学会94全国铁谱技术会议, 1994].
521(7553): 436–444. doi: 10.1038/nature14539.
[ 3 ] Wang Weihua, Yin Yonghui, Wang Chengtao. Wear debris
[13] Zheng Yuanpan, Li Guangyang, Li Ye. Survey of Application of
recognition system based on radius basis function network[J].
Deep Learning in Image Recognition[J]. Computer Engineering and
Tribology, 2003, (4): 77–80 (in Chinese) [王伟华, 殷勇辉, 王成焘. Applications, 2019, 55(12): 20–36 (in Chinese) [郑远攀, 李广阳, 李
基于径向基函数神经网络的磨粒识别系统[J]. 摩擦学学报, 2003, 晔. 深度学习在图像识别中的应用研究综述[J]. 计算机工程与应
(4): 77–80]. doi: 10.3321/j.issn:1004-0595.2003.04.017. 用, 2019, 55(12): 20–36]. doi: 10.3778/j.issn.1002-8331.1903-0031.
[ 4 ] Shi Hong, Zhang Shuai, Li Ang. Research on wear particle [14] An Chao, Wei Haijun, Liu Hong, et al. Intelligent identification of
recognition based on self-adapting support vector machine[J].
ferrographic wear particles based on convolution neural network[J].
Science Technology and Engineering, 2012, 12(32): 8543–8546, Modern Manufacturing Engineering, 2019, 7: 111–114 (in Chinese)
8552 (in Chinese) [石宏, 张帅, 李昂. 基于自适应支持向量机的磨 [安超, 魏海军, 刘竑, 等. 基于卷积神经网络的铁谱磨粒智能识别
粒识别技术研究[J]. 科学技术与工程, 2012, 12(32): 8543–8546, 研究[J]. 现代制造工程, 2019, 7: 111–114]. doi: 10.16731/j.cnki.1671-
8552]. doi: 10.3969/j.issn.1671-1815.2012.32.013. 3133.2019.07.018.
[ 5 ] Cui Hai, Kang Jianli. Research on wear debris recognition algorithm [15] Peng Y, Cai J, Wu T, et al. A hybrid convolutional neural network
based on IFNN[J]. Journal of Zhejiang Water Conservancy and for intelligent wear particle classification[J]. Tribology International,
Hydropower College, 2016, 28(3): 77–80 (in Chinese) [崔海, 康剑 2019, 138: 166–173. doi: 10.1016/j.triboint.2019.05.029.
莉. 基于片相似各项异性扩散的BP神经网络的磨粒识别研究[J]. [16] Wang S, Wu T H, Shao T, et al. Integrated model of BP neural
浙江水利水电学院学报, 2016, 28(3): 77–80]. doi: 10.3969/j.issn. network and CNN algorithm for automatic wear debris
1008-536X.2016.03.016. classification[J]. Wear, 2019, 426-427: 1761–1770. doi: 10.1016/
[ 6 ] Li Yanjun, Zuo Hongfu, Wu Zhenfeng. Failure of engine diagnostics j.wear.2018.12.087.