Abstract：Low- and medium-voltage cables with severe overload operation, harsh working environment and frequent overheating faults are often the source of high incidence of electrical fires, but there is a lack of economical and reliable overheating fault diagnosis methods in the early stages of fire. Gas component analysis has been applied to transformers (dissolved gas in oil detection method) and the diagnosis of overheating and discharge faults in SF6 and its alternative gas insulated equipment. In the early stage of fire, gas volatiles generally appear before smoke particles, and gas component analysis can provide faster early warning of fire than the traditional smoke detection method, so this paper carried out research on the cable overheating fault diagnosis method based on gas sensing array. Firstly, GC-MS was used to analyze the gas components of cable overheating decomposition. The gas decomposition components at different temperatures was compared, the number of gas components and the change of main component had been discussed. The result of GC-MS showed that the decomposition products of superheated PVC contains aromatic compounds, esters, aldehydes, ketones, saturated hydrocarbons, unsaturated hydrocarbons and other volatile organic compounds VOCs. With the increase of superheat temperature, the types of cable decomposition gas increased significantly. Among the decomposition gas, benzene, 2-EH, DOTP and DEHA were the four components with the highest gas concentrations. 2-EH exists in the whole temperature range of 90~200℃, DOTP and DEHA exist only at 140℃ and above, and the concentration of these three substances varies widely. Benzene mainly exists in the high temperature and serious overheating state above 180℃, and its content increases sharply at 200℃. Based on the above analysis results, a variety of commercial sensors for VOCs detectionwere selected to compose the gas sensor array. A cable overheating fault simulation experiment was conducted to record the response curves of the sensor array, and the cable was heated to different temperatures. As the cable superheat temperature increases, the gas sensor array response curve showed a trend of faster response rate and increased response amplitude. Most gas sensors reaching saturation in response within 5 min after the heating temperature stabilized.As a comparison, in the temperature range of 70~300℃, the smoke alarm never alarmed.The collected response data weretaken within 5 min after temperature stabilization, and 20 features wereextracted, including time domain features, frequency domain features, and curve fitting features.The effects of 4 feature dimensionality reduction methods such as principal component analysis,Select K Best, tree estimation, SelectFromModel, etc. werecompared and Select K Best worked best. The optimal feature dimensionality reduction method was selected and the features were redeuced from 20 dimensions to 10 dimensions. After this, the mapping relationship between cable overheating temperature and response features was established using machine learning algorithms. The classification accuracy of 8 machine learning algorithms such as Extra Tree, Random Forest, Decision Tree, K Near Neighbor, Logistic Regression, Naive Bayes, Adaptive Boosting, and AdaBoost were compared. SVM works best in the classifier on the test set, which can classify the cable operating temperature into 3 different states such as normal (temperature up to 90℃), warning (temperature 120~160℃) and alarm (temperature above 180℃), and the accuracy of the model reaches 95%. The conclusions of the above studies are as follows: (1) Cable superheat decomposition gas contains aromatic compounds, esters, aldehydes, ketones, saturated hydrocarbons, unsaturated hydrocarbons and other volatile organic compounds VOCs, so this paper selected a variety of commercial sensors for VOCs detection to form a sensor array. (2) As the cable superheat temperature increases, the response speed of the gas sensor response curve becomes faster and the response amplitude increases, and most gas sensors reach saturation in response within 5 min after the heating temperature stabilizes. In the temperature range of 70~300℃, the smoke detectordidn't alarm. (3) By extracting 20 features from the sensor array response signal within 5 minutes and filtering the first 10 optimal features according to the chi-square test and using SVM classification, the cable operating temperature can be classified into three states, such as normal (temperature up to 90℃), warning (temperature 120~160℃) and alarm (temperature higher than 180℃), and the accuracy of the model on test set reaches 95%. The cable overheating fault diagnosis method established in this paper has significant advantages over smoke alarms in terms of alarm timeliness as well as temperature interval identification accuracy.
雷芳菲, 褚继峰, 刘洋, 杨爱军, 袁欢, 王小华, 荣命哲. 基于半导体气体传感阵列的电缆过热故障诊断方法[J]. 电工技术学报, 2023, 38(13): 3651-3664.
Lei Fangfei, Chu Jifeng, Liu Yang, Yang Aijun, Yuan Huan, Wang Xiaohua, Rong Mingzhe. Fault Diagnosis of Cable Overheating Based on Semiconductor Gas Sensing Array. Transactions of China Electrotechnical Society, 2023, 38(13): 3651-3664.
[1] 王志. 线缆绝缘材料热解特性与线缆燃烧及火蔓延行为研究[D]. 合肥: 中国科学技术大学, 2020. [2] Liu Sanwei, Huang Fuyong, Gong Liang, et al.Research summary of cable channel fire extinguishing technology[J]. E3S Web of Conferences, 2021, 233: 04022. [3] 单秉亮, 李舒宁, 杨霄, 等. XLPE配电电缆缺陷诊断与定位技术面临的关键问题[J]. 电工技术学报, 2021, 36(22): 4809-4819. Shan Bingliang, Li Shuning, Yang Xiao, et al.Key problems faced by defect diagnosis and location technologies for XLPE distribution cables[J]. Transactions of China Electrotechnical Society, 2021, 36(22): 4809-4819. [4] 杜林, 余辉宗, 严涵. 基于构建温度场分析的非侵入式高压电缆缆芯温度测量方法[J]. 电工技术学报, 2021, 36(7): 1338-1346. Du Lin, Yu Huizong, Yan Han.A non-invasive temperature measurement method for high-voltage cable core based on temperature field construction and analysis[J]. Transactions of China Electrotechnical Society, 2021, 36(7): 1338-1346. [5] 陈志强, 王楠. 感温光纤检测装置的设计与实现[J]. 电工技术, 2021(14): 45-47. Chen Zhiqiang, Wang Nan.Design and implementation of temperature-sensing fiber inspecting device[J]. Electric Engineering, 2021(14): 45-47. [6] 董雄鹰. 基于分布式光纤测温的微电网电缆故障定位研究[D]. 北京: 华北电力大学(北京), 2021. [7] 江军, 马国明, 宋宏图, 等. 基于侧边抛磨光纤布拉格光栅的变压器油中溶解氢气传感器[J]. 电工技术学报, 2017, 32(13): 264-270. Jiang Jun, Ma Guoming, Song Hongtu, et al.Dissolved hydrogen sensor in power transformer oil based on side polishing fiber Bragg grating[J]. Transactions of China Electrotechnical Society, 2017, 32(13): 264-270. [8] 王鹏, 蔡贤卿. 测温技术在城市地下综合管廊中的研究与应用[J]. 城市建设理论研究(电子版), 2016(26): 98-99. Wang Peng, Cai Xianqing.Research and application of temperature measurement technology in urban underground utility tunnel[J]. Urban Construction Theory Research, 2016(26): 98-99. [9] 王秋翠, 魏立明. 城市地下管廊的火灾危险识别与预警系统研究[J]. 吉林建筑大学学报, 2019, 36(1): 82-86. Wang Qiucui, Wei Liming.Research on fire hazard identification and early warning system of urban underground pipe gallery[J]. Journal of Jilin Jianzhu University, 2019, 36(1): 82-86. [10] 吴科, 熊刚, 邓旭东, 等. 城市高压电缆分布式光纤测温技术应用现状[J]. 电工技术, 2020(1): 140-143. Wu Ke, Xiong Gang, Deng Xudong, et al.Application status of distributed temperature sensing technology for urban high voltage cables[J]. Electric Engineering, 2020(1): 140-143. [11] 常叶枫, 尚文迪, 张强. 基于LoRa的电缆无线测温装置设计与实现[J]. 电工技术, 2021(8): 66-68. Chang Yefeng, Shang Wendi, Zhang Qiang.Design and implementation of wireless temperature measurement device for power cable based on LoRa[J]. Electric Engineering, 2021(8): 66-68. [12] 林晨炯, 林珍, 吴雅琳. 电缆接头温度在线监测方法研究综述[J]. 电气技术, 2019, 20(5): 1-4, 9. Lin Chenjiong, Lin Zhen, Wu Yalin.Summary of research on online monitoring method of cable joint temperature[J]. Electrical Engineering, 2019, 20(5): 1-4, 9. [13] 王守相, 李皓哲, 刘琪, 等. 基于激磁电流单周期内负半波持续时间特征的直流剩余电流检测新型相位差法[J]. 电力系统保护与控制, 2021, 49(11): 45-53. Wang Shouxiang, Li Haozhe, Liu Qi, et al.A new phase difference method for DC residual current detection based on the characteristic of the duration of the exciting current negative half wave in a period[J]. Power System Protection and Control, 2021, 49(11): 45-53. [14] 张思祥, 甘凯, 周围. 提高火灾烟雾传感器检测精度的方法[J]. 传感器与微系统, 2021, 40(1): 151-153. Zhang Sixiang, Gan Kai, Zhou Wei.Method for detection precision improving of fire smog sensor[J]. Transducer and Microsystem Technologies, 2021, 40(1): 151-153. [15] Fonollosa J, Solórzano A, Marco S.Chemical sensor systems and associated algorithms for fire detection: a review[J]. Sensors (Basel, Switzerland), 2018, 18(2): 553. [16] Pannek C.Highly sensitive reflection based colorimetric gas sensor to detect CO in realistic fire scenarios[J]. Sensors and Actuators B: Chemical, 2020, 306: 127572. [17] Schmitt K, Tarantik K, Pannek C, et al.Colorimetric materials for fire gas detection—a review[J]. Chemosensors, 2018, 6(2): 14. [18] 杨天荷, 陈伟根, 李志军. 基于悬臂梁光纤F-P声传感器的光声光谱检测法及其对微量乙炔气体的检测[J]. 高电压技术, 2020, 46(6): 1922-1928. Yang Tianhe, Chen Weigen, Li Zhijun.Detection of trace ethyne by photoacoustic spectroscopy based on fiber-optic cantilever acoustic sensor[J]. High Voltage Engineering, 2020, 46(6): 1922-1928. [19] 张冬至, 姜传星, 殷乃良, 等. 纳米修饰石墨烯薄膜传感器的变压器油中乙炔气体敏感特性[J]. 高电压技术, 2017, 43(11): 3712-3717. Zhang Dongzhi, Jiang Chuanxing, Yin Nailiang, et al.Sensing characteristics of acetylene gas in transformer oil of nano-modified graphene-based film sensor[J]. High Voltage Engineering, 2017, 43(11): 3712-3717. [20] 张晓星, 刘恒, 张英, 等. 基于光声光谱峰面积的微量乙炔气体定量检测[J]. 高电压技术, 2015, 41(3): 857-863. Zhang Xiaoxing, Liu Heng, Zhang Ying, et al.Quantum detection of trace acetylene gas based on the peak area of photoacoustic spectroscopy[J]. High Voltage Engineering, 2015, 41(3): 857-863. [21] 王婧璇, 周渠, 桂银刚, 等. 掺杂硫化钼对油中特征气体C2H2的吸附性能[J]. 高电压技术, 2020, 46(6): 1962-1969. Wang Jingxuan, Zhou Qu, Gui Yingang, et al.Adsorption properties of the C2H2 characteristic gas in oil by doped MoS2[J]. High Voltage Engineering, 2020, 46(6): 1962-1969. [22] 陈图南, 马凤翔, 王刘芳, 等. 高分子渗透膜在变压器油中溶解气体分析中的应用[J]. 电工技术学报, 2022, 37(3): 750-766. Chen Tunan, Ma Fengxiang, Wang Liufang, et al.Application of polymers membrane in dissolved gas analysis: a review[J]. Transactions of China Electrotechnical Society, 2022, 37(3): 750-766. [23] 张燕, 方瑞明. 基于油中溶解气体动态网络标志物模型的变压器缺陷预警与辨识[J]. 电工技术学报, 2020, 35(9): 2032-2041. Zhang Yan, Fang Ruiming.Fault detection and identification of transformer based on dynamical network marker model of dissolved gas in oil[J]. Transactions of China Electrotechnical Society, 2020, 35(9): 2032-2041. [24] 马晨曦, 乔胜亚, 周文俊, 等. SF6变压器中悬浮电位放电分解气体的生成规律[J]. 高电压技术, 2020, 46(11): 4052-4060. Ma Chenxi, Qiao Shengya, Zhou Wenjun, et al.Decomposed gases generated by partial discharges at floating potential defects in SF6 gas transformers[J]. High Voltage Engineering, 2020, 46(11): 4052-4060. [25] 张英, 张晓星, 李军卫, 等. 基于光声光谱法的SF6气体分解组分在线监测技术[J]. 高电压技术, 2016, 42(9): 2995-3002. Zhang Ying, Zhang Xiaoxing, Li Junwei, et al.On-line monitoring technology of SF6 gas decomposition components based on photoacoustic spectroscopy[J]. High Voltage Engineering, 2016, 42(9): 2995-3002. [26] 周朕蕊, 韩冬, 赵明月, 等. 电晕放电下C5F10O混合气体的分解特性[J]. 电工技术学报, 2021, 36(2): 407-416. Zhou Zhenrui, Han Dong, Zhao Mingyue, et al.Decomposition characteristics of C5F10O mixtures under corona discharge[J]. Transactions of China Electrotechnical Society, 2021, 36(2): 407-416. [27] Kumar R, Liu Xianghong, Zhang Jun, et al.Room-temperature gas sensors under photoactivation: from metal oxides to 2D materials[J]. Nano-Micro Letters, 2020, 12(1): 164. [28] 张晓星, 方佳妮, 崔豪, 等. 碳纳米管传感器吸附机理及对SF6分解组分检测应用综述[J]. 中国电机工程学报, 2018, 38(16): 4926-4941, 4997. Zhang Xiaoxing, Fang Jiani, Cui Hao, et al.Review: adsorption principle and application of carbon nanotubes to SF6 decomposition components[J]. Proceedings of the CSEE, 2018, 38(16): 4926-4941, 4997. [29] Chu Jifeng, Yang Xu, Yang Aijun, et al.Multivariate evaluation method for screening optimum gas-sensitive materials for detecting SF6 decomposition products[J]. ACS Sensors, 2020, 5(7): 2025-2035. [30] Chu Jifeng, Li Weijuan, Yang Xu, et al.Identification of gas mixtures via sensor array combining with neural networks[J]. Sensors and Actuators B: Chemical, 2021, 329: 129090. [31] Beneš M, Milanov N, Matuschek G, et al.Thermal degradation of PVC cable insulation studied by simultaneous TG-FTIR and TG-EGA methods[J]. Journal of Thermal Analysis and Calorimetry, 2004, 78(2): 621-630. [32] Zhou Jingjing.TG-FTIR and Py-GC/MS study of the pyrolysis mechanism and composition of volatiles from flash pyrolysis of PVC[J]. Journal of the Energy Institute, 2020, 93(6): 2362-2370. [33] 魏艳慧, 郑元浩, 龙海泳, 等. 绝缘层厚度对高压直流电缆电场和温度场分布的影响[J]. 电工技术学报, 2022, 37(15): 3932-3940. Wei Yanhui, Zheng Yuanhao, Long Haiyong, et al.Influence of insulation layer thickness on electric field and temperature field of HVDC cable[J]. Transactions of China Electrotechnical Society, 2022, 37(15): 3932-3940. [34] Wang Xiaohua, Wang Dawei, Yang Aijun, et al.Effects of adatom and gas molecule adsorption on the physical properties of tellurene: a first principles investigation[J]. Physical Chemistry Chemical Physics: PCCP, 2018, 20(6): 4058-4066. [35] 桂银刚, 许文龙, 张晓星, 等. TiO2掺杂石墨烯对SO2气体的气敏特性研究[J]. 电工技术学报, 2021, 36(21): 4590-4597. Gui Yingang, Xu Wenlong, Zhang Xiaoxing, et al.Adsorption property of SO2 gas on TiO2-doped graphene[J]. Transactions of China Electrotechnical Society, 2021, 36(21): 4590-4597. [36] Zhang Hao, Feng Jianchao, Fei Teng, et al.SnO2 nanoparticles-reduced graphene oxide nanocomposites for NO2 sensing at low operating temperature[J]. Sensors and Actuators B: Chemical, 2014, 190: 472-478. [37] Yang Aijun, Wang Dawei, Lan Tiansong, et al.Single ultrathin WO3 nanowire as a superior gas sensor for SO2 and H2S: Selective adsorption and distinct I-V response[J]. Materials Chemistry and Physics, 2020, 240: 122165. [38] 国家质量监督检验检疫总局, 中国国家标准化管理委员会. GB/T 5023.1—2008 额定电压450/750V及以下聚氯乙烯绝缘电缆第1部分:一般要求[S]. 北京: 中国标准出版社, 2009. [39] 国家质量监督检验检疫总局、中国国家标准化管理委员会. GB/T 8815—2008 电线电缆用软聚氯乙烯塑料[S]. 北京: 中国标准出版社, 2009. [40] Yan Jia, Guo Xiuzhen, Duan Shukai, et al.Electronic nose feature extraction methods: a review[J]. Sensors (Basel, Switzerland), 2015, 15(11): 27804-27831. [41] 闫嘉. 基于电子鼻技术的人体伤口感染检测系统构建及算法研究[D]. 重庆: 重庆大学, 2012. [42] Xu Sen, Hua Xiaopeng, Xu Jing,et al.Cluster ensemble approach based on T-distributed stochastic neighbor embedding[J]. Journal of Electronics & Information Technology, 2018, 40(6): 1316-1322. [43] 王昱皓, 武建文, 马速良, 等. 基于核主成分分析-SoftMax的高压断路器机械故障诊断技术研究[J]. 电工技术学报, 2020, 35(增刊1): 267-276. Wang Yuhao, Wu Jianwen, Ma Suliang, et al.Mechanical fault diagnosis research of high voltage circuit breaker based on kernel principal component analysis and SoftMax[J]. Transactions of China Electrotechnical Society, 2020, 35(S1): 267-276. [44] 马速良, 武建文, 袁洋, 等. 多振动信息下的高压断路器机械故障随机森林融合诊断方法[J]. 电工技术学报, 2020, 35(增刊2): 421-431. Ma Suliang, Wu Jianwen, Yuan Yang, et al.Mechanical fault fusion diagnosis of high voltage circuit breaker using multi-vibration information based on random forest[J]. Transactions of China Electrotechnical Society, 2020, 35(S2): 421-431. [45] Liu Zhi, Wang Liyang, Zhang Yun, et al.A SVM controller for the stable walking of biped robots based on small sample sizes[J]. Applied Soft Computing, 2016, 38: 738-753. [46] 赵书涛, 许文杰, 刘会兰, 等. 基于振动信号谱形状熵特征的高压断路器操动状态辨识方法[J]. 电工技术学报, 2022, 37(9): 2170-2178. Zhao Shutao, Xu Wenjie, Liu Huilan, et al.Identification method for operation state of high voltage circuit breakers based on spectral shape entropy characteristics of vibration signals[J]. Transactions of China Electrotechnical Society, 2022, 37(9): 2170-2178. [47] 炜盛科技. 气体传感器[EB/OL]. [2023-03-02].https://www.winsensor.com/product/. [48] 费加罗传感科技. 气体传感器[EB/OL]. [2022-05-28]. http://www.figaro-china.com/product/sensor/.
2023, Vol. 38 
