基于机器学习与红外光谱技术的变压器油老化行为研究

doi:10.3964/j.issn.1000-0593(2025)02-0434-09

摘要
参考文献
相关文章 (15)

全文: PDF (17784 KB)
输出: BibTeX | EndNote (RIS)

摘要：为解决现阶段油品老化分析工作复杂、误差大等问题，提出一种红外光谱与机器学习（ML）相融合的技术。借助傅里叶变换中红外（FT-MIR）光谱仪采集三种变压器油在不同老化时间的样本光谱，运用多种预处理方法对样本光谱进行预处理，以自动寻峰并求得特征峰面积之和。采用偏最小二乘回归（PLSR）和粒子群优化-支持向量机回归（PSO-SVR）算法建立了变压器油老化程度定量分析模型，研究并分析了多种光谱数据预处理方法对红外光谱降噪、基线校正等处理效果以及对两种模型定量分析效果的影响。结果表明，油品光谱预处理效果最好的是平滑法，其中SG+SVR和SG+PLSR模型拟合优度（R²）分别为0.981 4、0.991 3，平均绝对误差（MAE）为0.312 4、0.288 0，均方根误差（RMSE）仅有0.097 7、0.379 0。在合适的预处理条件下，两种机器学习算法鲁棒性和可靠性均较强，模型预测值与实际值间差异极小。

关键词：机器学习；傅里叶变换中红外光谱；变压器油；老化程度；粒子群优化-支持向量机回归(PSO-SVR)；偏最小二乘回归(PLSR)

Abstract：To solve the problems of complexity and large errors in oil aging analysis at the present stage, a technique integrating infrared spectroscopy and machine learning is proposed. With the help of a Fourier-Transform Mid-Infrared (FT-MIR) spectrometer, the sample spectra of three kinds of transformer oils were collected at different aging times. Various preprocessing methods were used to preprocess the sample spectra, and then the peaks were automatically sought and the sum of the characteristic peak areas was obtained. PLSR and PSO-SVR were used to establish a quantitative analysis model of transformer oil aging degree, and the effects of multiple spectral data preprocessing methods on the processing effects of infrared spectral noise reduction and baseline correction, as well as on the quantitative analysis effects of two models were investigated and analyzed. The results show that the best oil spectral preprocessing is the smoothing method, in which the SG+SVR and SG+PLSR model fitting Goodness-of-Fit (R²) are 98.14% and 99.13%, respectively, and the mean absolute error (MAE) is 0.312 4 and 0.288 0, and the root-mean-square error(RMSE) is only 0.097 7 and 0.379 0. Under the appropriate preprocessing conditions, both machine learning algorithms are robust and reliable, and the difference between the predicted and actual values of the models is extremely small.

Key words：Machine learning; Fourier-transform mid-infrared-spectroscopy; Transformer oil; Degree of aging; PSO-SVR;PLSR

收稿日期: 2024-01-14 修订日期: 2024-05-28

中图分类号:

O657.33

基金资助: 国家自然科学基金项目（21501015, 31527803, 21545010），湖南省自然科学基金项目（2022JJ30604）资助

通讯作者: 宋刘斌 E-mail: liubinsong1981@126.com

作者简介: 肖忠良，1964年生，长沙理工大学化学化工学院教授 e-mail: xiaozhongliang@163.com

引用本文:

肖忠良，袁荣耀，付壮，刘成，尹碧露，肖敏之，赵亭亭，匡尹杰，宋刘斌. 基于机器学习与红外光谱技术的变压器油老化行为研究[J]. 光谱学与光谱分析, 2025, 45(02): 434-442.
XIAO Zhong-liang, YUAN Rong-yao, FU Zhuang, LIU Cheng, YIN Bi-lu, XIAO Min-zhi, ZHAO Ting-ting, KUANG Yin-jie, SONG Liu-bin. Study on the Aging Behavior of Transformer Oil Based on Machine Learning and Infrared Spectroscopy Technology. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2025, 45(02): 434-442.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2025)02-0434-09 或 https://www.gpxygpfx.com/CN/Y2025/V45/I02/434

[1] Ugochukwu E，Azam N，Arshad A, et al. Sensors，2022，22(20): 7923.
[2] Kalashnikov D A，Paterova A V，Kulik S P, et al. Nature Photonics，2016，10(2): 98.
[3] Zhang W，Kasun L C，Wang Q J, et al. Sensors，2022，22(24): 9764.
[4] CHEN Lu, CHANG Long-fei, SHEN Mu-ao, et al(陈露，常龙飞，沈沐傲, 等). Journal of Instrumental Analysis(分析测试学报)，2024，43(3): 489.
[5] ZHANG Hai-liang, XIE Chao-yong, TIAN Peng, et al(章海亮，谢潮勇，田彭, 等). Spectroscopy and Spectral Analysis(光谱学与光谱分析)，2023，43(7): 2226.
[6] YANG Cheng-en, SU Ling, FENG Wei-zhi, et al(杨承恩，苏玲，冯伟志, 等). Spectroscopy and Spectral Analysis(光谱学与光谱分析)，2023，43(2): 577.
[7] ZHU Ya-kun, LIU Ru-can, ZHANG Er-qiang, et al(朱亚昆，刘如灿，张二强, 等). Tobaco Science & Technology(烟草科技)，2024，57(6): 7.
[8] REN Shuang-zan, ZHONG Li-sheng, YU Qin-xue, et al(任双赞，钟力生，于钦学, 等). Journal of Xi'an Jiaotong University(西安交通大学学报)，2010，44(10): 88.
[9] Abdelrahman M Alshehawy, Diaa-Eldin A Mansour，Mohsen Ghali, et al. Processes，2021，9(5): 732.
[10] Yuan B，Wang L，Liu P, et al. Genetics in Medicine，2020，22(10): 1633.
[11] Hong J H，Hwang E S，McManus M T, et al. Science，2005，309(5737): 1074.
[12] Hailing L，Chaohua Y，Linchun W, et al. Physica A: Statistical Mechanics and Its Applications, 2022, 607: 128227.
[13] Clifford C Shekter, Day Yi, Hina J Panchal, et al. Plastic and Reconstructive Surgery，2018，142(3): 493.
[14] Mohammed Ayman R, Hassan Kumail S, et al. IFAC PapersOnLine，2022，55(10): 749.
[15] Yi H，Fangcheng K，Xin W, et al. Building and Environment，2022，211: 108723.
[16] Simone P，Amanda D d M，Edoardo L, et al. Molecules，2021，26(20): 6245.
[17] Yihe W，Dave Z S. Computational Statistics & Data Analysis，2021，156: 107130.
[18] Jokin E，Daniel S-G，Laura A, et al. Chemometrics and Intelligent Laboratory Systems，2023, 240: 104883.
[19] Qian Y，Carolina B M，Eleanor S, et al. PLoS Medicine，2022，19(9): e1004090.
[20] GUO Peng, ZHAO Yang, SUN Zi-hao, et al(郭鹏，赵阳，孙子皓, 等). China Agricultural Informatics(中国农业信息)，2023，35(1): 55.
[21] Bing Cheng Z，Hua Min L，Shao Hui L, et al. British Journal of Anaesthesia, 2022, 129(5): e140-e142.
[22] Yizhen H，Lin H，Xiaoqing Y, et al. Food Chemistry: X, 2023, 19: 100733.
[23] Yamina S，Farid Y, Abdelhamid I. Electric Power Systems Research, 2022, 210: 108153.
[24] Chen J，Ma S, Wu Y. Journal of Ambient Intelligence and Humanized Computing, 2022, 13: 5699.
[25] Wenying Z，Huaguang Z，Jinhai L, et al. IEEE/CAA Journal of Automatica Sinica, 2017, 4(3): 520.
[26] Jie Shi，Wei Jen Lee, Yongqiang Liu, et al. IEEE Transactions on Industry Applications, 2012, 48(3): 1064.
[27] Gupta D，Borah P，Sharma U M, et al. Neural Computing and Applications, 2022, 34: 11335.
[28] Mengwei S，Prayag T，Yuqin Q, et al. Knowledge-Based Systems, 2022, 250: 109174.
[29] Eun P J，H N V，Pei-Chien T, et al. Gastroenterology, 2023, 164(6S).
[30] Mishra P，Karami A，Nordon A, et al. Sensors Actuators B: Chemical, 2019, 281: 1034.