基于FCOS神经网络的制动主缸内槽缺陷检测方法

王芷薇; 郭斌; 胡晓峰; 罗哉; 段林茂

doi:10.3969/j.issn.1000-1158.2021.09.17

PDF(2819 KB)

计量学报 ›› 2021, Vol. 42 ›› Issue (9) : 1225-1231. DOI: 10.3969/j.issn.1000-1158.2021.09.17

无线电、时间频率计量

基于FCOS神经网络的制动主缸内槽缺陷检测方法

王芷薇¹,郭斌¹,胡晓峰^1,2,罗哉¹,段林茂³

作者信息 +

Internal Groove Defect Detection Method of Brake Master Cylinder Based on FCOS Neural Network

WANG Zhi-wei¹,GUO Bin¹,HU Xiao-feng^1,2,LUO Zai¹,DUAN Lin-mao³

Author information +

文章历史 +

摘要

针对主缸内槽缺陷检测存在干扰因素复杂、检测精度低等难点,提出了一种基于全卷积单阶段神经网络(FCOS)的主缸内槽缺陷检测算法。利用特征融合金字塔网络进行特征提取并逐像素预测,得到缺陷种类,实现凹槽缺陷的自动检测。实验结果表明,FCOS网络对制动主缸内槽砂眼、划痕、振刀纹缺陷检测的平均精度均值分别为85.2％、87.5％、90.1％,精确度分别为0.98、0.89、0.95。实验结果与Mask R-CNN网络和Faster R-CNN网络的实验结果进行对比,FCOS网络具有更高的准确率,学习时长大幅度缩短,且满足实时检测要求。

Abstract

Aiming at the difficulties of complicated interference factors and low detection accuracy in the detection of groove defects in the main cylinder, a detection algorithm for groove defects in the main cylinder based on full convolution single stage neural network (FCOS) was proposed. FPN network was used for feature extraction and pixel by pixel prediction, and the predicted results were classified to realize automatic detection of groove defects. The experimental results show that the mAP values of FCOS network in detecting the sand hole, scratch and vibration pattern in the inner groove of the main cylinder are 85.2%, 87.5% and 90.1%,and the detection accuracy is 0.98, 0.89 and 0.95. Finally, the experimental results were compared with those of the Mask R-CNN network and Faster R-CNN network. FCOS network had higher accuracy, significantly shortened learning time and satisfied real-time detection requirements.

导出引用

王芷薇,郭斌,胡晓峰,罗哉,段林茂. 基于FCOS神经网络的制动主缸内槽缺陷检测方法[J]. 计量学报. 2021, 42(9): 1225-1231 https://doi.org/10.3969/j.issn.1000-1158.2021.09.17

WANG Zhi-wei,GUO Bin,HU Xiao-feng,LUO Zai,DUAN Lin-mao. Internal Groove Defect Detection Method of Brake Master Cylinder Based on FCOS Neural Network[J]. Acta Metrologica Sinica. 2021, 42(9): 1225-1231 https://doi.org/10.3969/j.issn.1000-1158.2021.09.17

中图分类号： TB973

参考文献

［1］张晋, 孔祥东, 姚静, 等. 汽车防抱死制动系统液压控制单元的建模与仿真［J］. 中国机械工程, 2016, 27(21): 2967-2974.
Zhang J, Kong X D, Yao J, et al. Modeling and simulation of ABS HCU［J］. China Mechanical Engineering, 2016, 27(21): 2967-2974.
［2］胡晓峰,江文松,郭斌, 等. 串联式制动阀静特性仿真及测试方法研究［J］. 计量学报, 2019, 40(6): 1088-1095.
Hu X F, Jiang W S, Guo B, et al. Study On the Simulation and Test Method of Static Characteristics for Series Brake Valve［J］. Acta Metrologica Sinica, 2019, 40(6): 1088-1095.
［3］高俊鹏, 姜涛, 张桂林, 等. 汽车制动主缸补偿孔形位尺寸检测双远心光学系统研究［J］. 计量学报, 2017, 38(3): 262-266.
Gao J P, Jiang T, Zhang G L, et al.Research on Double Telecentric Optical System for the Form and Position Detection of Automobile Brake Cylinder Comoensation Hole［J］. Acta Metrologica Sinica, 2017, 38(3): 262-266.
［4］郭斌, 华士丹, 范伟军, 等. 柱塞主缸密封凹槽表面缺陷检测系统［J］. 中国测试, 2018, 44(8): 76-81.
Guo B, Hua S D, Fan W J, et al.Surface Defect Detection System for Sealing Groove of Plunger Main Cylinder［J］. China Test, 2018, 44(8): 76-81.
［5］石韵昭, 倪军, 范晟华, 等. 基于机器视觉的双工位载带检测系统研究［J］. 计量学报, 2019, 40(2): 196-200.
Shi Y Z, Ni J, Fan S H, et al.Research on dual station carrier-stripetection system based on machine vision［J］. Acta Metrologica Sinica, 2019, 40(2): 196-200.
［6］程瑶, 赵雷, 成珊, 等. 基于机器视觉的车距检测系统设计［J］. 计量学报, 2020, 41(1): 11-15.
Cheng Y, Zhao L, Cheng S, et al.Design of vehicledistanced-这个-是为了调版加的etectionsystem based on machine vision［J］. Acta Metrologica Sinica, 2020, 41(1): 11-15.
［7］晏世武, 罗金良, 严庆. 基于改进Yolov3的目标检测的研究［J］. 智能计算机与应用, 2019, 9(6): 312-315.
Yan S W, Luo J L, Yan Q. Research on target detection based on improved Yolov3［J］. Intelligent computers and applications, 2019, 9(6): 312-315.
［8］李小宁, 雷涛, 钟剑丹, 等. 基于改进SSD的车辆小目标检测方法［J］. 应用光学, 2020, 41(1): 150-155.
Li X N, Lei T, Zhong J D, et al.Vehicle small target detection method based on improved SSD ［J］. Applied optics, 2020, 41(1): 150-155.
［9］Ahmed Selsayed, Hala Mebeid, Mohamed Roushdy. Robust palm and knuckle ROI extraction in unconstrained environment［J］. Pattern Analysis and Applications, 2019, 22(4), 1537-1559.
［10］Bleasby J. ROI analysis key before new tech investment［J］. Daily Commercial News, 2019, 92(211): 1-2.
［11］侯志强, 刘晓义, 余旺盛, 等. 基于双阈值-非极大值抑制的Faster R-CNN改进算法［J］. 光电工程, 2019, 46(12): 82-92.
Hou Z Q, Liu X Y, Yu W S, et al.Improved Faster R-CNN Algorithm Based on Double Threshold-Non-Maximum Suppression［J］. Optoelectronic Component, 2019, 46(12): 82-92.
［12］刘庆民, 张蕾, 吴立群, 等. 基于机器视觉的非均匀分布点圆度误差评定［J］. 计量学报, 2016, 37(6): 567-570.
Liu Q M, Zhang L, Wu L Q, et al.Evaluation of roundness error of inhomogeneous distribution points based on machine vision［J］. Acta Metrologica Sinica, 2016, 37(6): 567-570.
［13］李国友, 纪执安, 张凤煦. 融合多层深度特征的核相关滤波跟踪算法［J］. 高技术通讯, 2020, 30(2): 126-133.
Li G Y, Ji Z A, Zhang F X. A kernel correlation filtertracking algorithm based on multi-layer depthfeature［J］. High technology communication, 2020, 30(2): 126-133.
［14］李良福, 宋睿, 冯建云, 等. 基于深度降噪自编码器的多特征目标融合跟踪算法［J］. 光电子·激光, 2020, 31(2): 175-186.
Li L F, Song R, Feng J Y, et al.A Multi-feature target fusion tracking algorithm based on deep de-noising self-encoder［J］. Photoelectron lase, 2020, 31(2): 175-186.
［15］徐亮, 符冉迪, 金炜, 等. 基于多尺度特征损失函数的图像超分辨率重建［J］. 光电工程, 2019, 46(11): 3-11.
Xu L, Fu R D, Jin W, et al.Image super resolution reconstruction based on multi-scale feature loss function［J］. OptoelectronicComponent, 2019, 46(11): 3-11.
［16］郭保苏, 吴文文, 付强, 等. 基于支持向量机分类策略的多晶硅电池片色差检测［J］. 计量学报, 2019, 40(6): 1013-1019.
Guo B S, Wu W W, Fu Q, et al.Face Chromatic aberration detection of polysilicon cells based on support vector machine classification strategy［J］. Acta Metrologica Sinica, 2019, 40(6): 1013-1019.
［17］赵一中, 刘文波. 基于深度信念网络的非限制性人脸识别算法研究［J］. 计量学报, 2017, 38(1): 65-68.
Zhao Y Z, Liu W B. Research on unrestricted face recognition algorithm based on deep belief network［J］. Acta Metrologica Sinica, 2017, 38(1): 65-68.