[1]侯作云.doi: 10.3969/j.issn.1001-3849.2026.05.014基于太赫兹技术与TLBO算法的汽车电镀涂层厚度检测[J].电镀与精饰,2026,(05):95-104.
　HOU Zuoyun.Thickness detection of electroplated automotive coating using terahertz technology and TLBO algorithm[J].Plating & Finishing,2026,(05):95-104.
点击复制

doi: 10.3969/j.issn.1001-3849.2026.05.014基于太赫兹技术与TLBO算法的汽车电镀涂层厚度检测()

《电镀与精饰》[ISSN:1001-3849/CN:12-1096/TG]

卷:
期数:
2026年05
页码:
95-104
栏目:
出版日期:
2026-05-31

文章信息/Info

Title:
Thickness detection of electroplated automotive coating using terahertz technology and TLBO algorithm
作者:
侯作云
(信阳艺术职业学院 公共教学部 河南 信阳 464000)
Author(s):
HOU Zuoyun
(Public Teaching Department, Xinyang Vocational College of Art, Xinyang 464000, China)
关键词:
太赫兹; 教与学优化算法(TLBO); Elman神经网络; 无损检测
Keywords:
terahertz;  teaching-learning-based optimization (TLBO);  Elman neural network;  non-destructive testing
分类号:
TQ050.9；TG17；U466
文献标志码:
A
摘要:
针对汽车电镀涂层厚度检测中工业噪声干扰与实时性难以协同的难题，提出太赫兹动态核宽滤波与群体智能优化神经网络融合架构。通过噪声能量驱动的高斯核宽自适应调节机制动态适配工业噪声强度，结合主成分分析压缩时频特征矩阵以消除计算冗余。采用双阶段教与学优化策略，教学阶段由精英个体引导动态教学强度，学习阶段通过适应度差异控制协作更新。此外，研究设计的教与学优化Elman网络，利用隐层状态反馈建模厚度时序依赖特性，显著提升检测鲁棒性。实验表明：该方法在5类基体上实现了0.65 μm平均绝对误差与52.0 dB峰值信噪比，特征提取延迟11.5 ms，工业误检率均值为1.9%；单次检测能耗23.3 mJ，内存占用峰值9.8 MB，边缘部署能耗波动标准差为1.5 mJ。该架构将镀层混叠工况检测性能提升了13%，为多材质复杂曲面镀层提供高精度厚度检测方案。
Abstract:
Aiming at the problem that industrial noise interference and real-time performance are difficult to coordinate in the thickness detection of electroplated automotive coatings, a fusion architecture of terahertz dynamic kernel width filtering and swarm intelligence optimized neural network were studied and proposed. The intensity of industrial noise is dynamically adapted to through a Gaussian kernel width adaptive adjustment mechanism driven by noise energy, and principal component analysis is combined to compress the time-frequency feature matrix to eliminate computational redundancy. A two-stage teaching and learning optimization strategy is adopted. In the teaching stage, elites guide the dynamic teaching intensity, and in the learning stage, fitness differences control collaborative updates. In addition, the teaching and learning optimization (TLBO) Elman network designed in the research utilizes hidden layer state feedback to model the thickness temporal dependence characteristics, significantly enhancing the detection robustness. Results show that this method achieves an average absolute error of 0.65 μm and a peak signal-to-noise ratio of 52.0 dB on five types of matrices, with a feature extraction delay of 11.5 ms and an average industrial false detection rate of 1.9%. The energy consumption for a single detection is 23.3 mJ, the peak memory usage is 9.8 MB, and the standard deviation of energy consumption fluctuation for edge deployment is 1.5 mJ. Moreover, this architecture enhances the detection performance of coating aliasing conditions by 13%, providing a high-precision thickness detection solution for multi-material complex curved surface coatings

参考文献/References:

[1].蒋智慧, 马迪卓, 赵艳亮. 汽车用车身材料不同油漆前处理微观结构及防腐能力[J]. 电镀与涂饰, 2022, 41(10): 718-722.
[2].LI C, FAN M, CAO B, et al. Thickness measurement of thermal barrier coating based on mutual inductance of eddy current system[J]. IEEE Transactions on Industrial Electronics, 2023, 71(7): 8099-8109.
[3].SINGH S A, DESAI K A. Automated surface defect detection framework using machine vision and convolutional neural networks[J]. Journal of Intelligent Manufacturing, 2023, 34(4): 1995-2011.
[4].李海港, 黄宇蕾, 朱美强, 等. 基于SPGL1算法的CFRP涂层厚度检测方法研究[J]. 实验技术与管理, 2025, 42(4): 127-135.
[5].XIE Y, HU W, XIE S, et al. Surface defect detection algorithm based on feature-enhanced YOLO[J]. Cognitive Computation, 2023, 15(2): 565-579.
[6].林滨, 杜安尧, 刘再蔚, 等. 基于分数阶傅里叶变换的声表面波涂层厚度检测[J]. 天津大学学报, 2022, 55(7): 714-720.
[7].NATEGH K. Improving the nondestructive thermography inspection results for detection of circular defects in coated metals using principal component analysis[J]. NDT Technology, 2022, 2(9): 33-40.
[8].SELVARAJAN L, VENKATARAMANAN K, CHOUDHURY B, et al. Evaluating of surface morphology on EDD machining parameters for Si 3N4-TiN using TLBO-MRA[J]. Materials and Manufacturing Processes, 2024, 39(4): 465-479.
[9].DUTTA S, SARMA D K, BARUA B M. Measurement and evaluation of coating performance characteristics of Hastelloy C276 processed by WS2 lubricant added meso-EDA operation[J]. Measurement, 2025: 118426.
[10].万玉红, 董形影, 吴育衡, 等. 太赫兹时域光谱技术在涂层检测中的研究进展[J]. 测控技术, 2023, 42(6): 22-35.
[11].曹绍林, 蔡煜, 唐伟军, 等. 基于TLBO的工程结构表面缺陷图像边缘检测方法[J]. 广州建筑, 2023, 51(6): 114-117.
[12].TU W, ZHONG S, ZHANG Q, et al. Rapid diagnosis of corrosion beneath epoxy protective coating using non-contact THz-TDS technique[J]. Nondestructive Testing and Evaluation, 2024, 39(3): 557-572.
[13].TANG H, HAN H, ZHANG S, et al. A generalized t-distribution-based kernel adaptive filtering algorithm[J]. IEEE Transactions on Circuits and Systems II: Express Briefs, 2024, 71(6): 3241-3245.
[14].王一鸣, 王彩玲, 王洪伟. 基于PCA透射光谱重构降噪的水体BOD含量模拟估算[J]. 光谱学与光谱分析, 2025, 45(2): 386-393.
[15].陈怡君, 任春年, 党妍洁, 等. 一种带有附加记忆策略的改进教与学优化算法[J]. 计算机技术与发展, 2023, 33(9): 208-214.
[16].RAJ P, BIJU P L, DEEPANRAJ B, et al. Optimizing the machining conditions in turning hybrid aluminium nanocomposites adopting teaching–learning based optimization and MOORA technique[J]. International Journal on Interactive Design and Manufacturing, 2024, 18(5): 3089-3101.
[17].翟志波, 贾国平, 王涛, 等. 基于拉普拉斯分布与鲍德温效应的教与学算法及其应用[J]. 计算机集成制造系统, 2023, 29(8): 2611-2621.
[18].刘潇, 孙世政, 张辉, 等. 基于粒子群优化Elman神经网络的流量温度复合测量[J]. 光电子?激光, 2024, 35(11): 1183-1191.
[19].LI C, WANG Y, CHEN X, et al. Elman neural network-based temperature prediction and optimization for lithium-ion batteries[J]. Proceedings of the Institution of Mechanical Engineers, Part D: Journal of Automobile Engineering, 2024, 238(10-11): 3448-3465.

更新日期/Last Update: 2026-05-12