A Non-contact Alcohol Measurement Method Based on Neural Network Correction Algorithm
-
摘要: 为了解决酒精气体测量过程中其他外界因素对测量浓度影响的问题,本文结合酒精气体在红外谱段吸收的特性以及BP神经网络算法的非线性处理方法提出了一种基于神经网络校正算法的酒精气体非接触测量方法。该算法考虑气体吸收过程中温度、湿度对光强的影响,把其作为神经网络的输入和测量参数一起进行训练,同时与常规的数据拟合模型算法进行对比实验,实验证明该算法取得了较好的效果。Abstract: This paper presents a non-contact method for the measurement of alcohol gas emission based on the neural network correction algorithm, to mitigate the influence of external factors on the measurement process. The proposed method combines the characteristics of alcohol gas absorption in the infrared spectrum and the nonlinear processing method of the back propagation(BP) neural network algorithm. The algorithm considers the influence of temperature and humidity on light intensity during the gas absorption process and trains it as the input to the neural network and measurement parameters. Simultaneously, the proposed algorithm is compared with the data fitting algorithm, and the experimental results show that this algorithm achieves better results.
-
Key words:
- infrared absorption /
- exponential fitting /
- BP neural network /
- alcohol gas /
- non-contact
-
表 1 酒精浓度-电压表
Table 1. Table of alcohol concentration and voltmeter
Alcohol concentration/ppm Signal voltage/ Reference voltage 60 1.025 80 1.0245 100 1.023 120 1.0212 140 1.020 160 1.018 180 1.0175 200 1.0165 表 2 模型评价参数表
Table 2. Table of model evaluation parameter
Evaluation parameters Sum of squares due to error R-square Adjusted R-square Root mean square error Value 9.169×10-7 0.9876 0.9784 0.0004788 表 3 部分预测误差表
Table 3. Table of partial prediction error
Standard gas concentration Prediction concentration of exponential fitting method Prediction concentration of BP neural network method Prediction error of exponential fitting method Prediction error of BP neural network method 100 91.7 96.1 8.266669052 3.920268702 100 105.4 91.6 -5.438678342 8.427306293 100 94.1 91.7 5.886509571 8.315740568 100 97.8 101.8 2.234567379 -1.766418361 100 101.0 94.4 -1.035570639 5.626445231 100 94.6 99.1 5.42093496 0.946998419 100 102.8 104.5 -2.838812408 -4.466340008 100 99.7 96.5 0.310392552 3.51272509 100 93.0 100.4 6.963089498 -0.421282919 100 105.6 101.4 -5.638639332 -1.371489129 100 92.0 96.5 7.987873553 3.510365108 100 95.9 99.8 4.118673325 0.167529403 100 94.7 91.6 5.252539606 8.353674239 表 4 误差标准差
Table 4. Standard deviation of error
Method Standard deviation of error Exponential fitting method 4.85271647048 BP neural network method 4.47312610406 -
[1] 王志芳, 王书涛, 王贵川. 粒子群优化BP神经网络在甲烷检测中的应用[J]. 光子学报, 2019, 48(4): 147-154. https://www.cnki.com.cn/Article/CJFDTOTAL-GZXB201904020.htmWANG Zhifang, WANG Shutao, WANG Guichuan. Application of particle swarm optimization BP neural network in methane detection[J]. Acta Photonica Sinica, 2019, 48(4): 147-154. https://www.cnki.com.cn/Article/CJFDTOTAL-GZXB201904020.htm [2] 童敏明, 张愉, 齐美星. 基于BP神经网络的可燃混合气体分析方法的研究[J]. 计量学报, 2006, 27(2): 169-171. doi: 10.3321/j.issn:1000-1158.2006.02.019TONG Minming, ZHANG Yu, QI Meixing. Study on analysis method of combustible gas mixture based on BP neural network[J]. Acta Metrologica Sinica, 2006, 27(2): 169-171. doi: 10.3321/j.issn:1000-1158.2006.02.019 [3] 太惠玲, 谢光忠, 蒋亚东. 基于气体传感器阵列的混合气体定量分析[J]. 仪器仪表学报, 2006, 27(7): 666-670. doi: 10.3321/j.issn:0254-3087.2006.07.002TAI Huiling, XIE Guangzhong, JIANG Yadong. Quantitative analysis of mixed gas based on gas sensor array[J]. Chinese Journal of Scientific Instrument, 2006, 27(7): 666-670 doi: 10.3321/j.issn:0254-3087.2006.07.002 [4] Philip Antwi, LI Jianzheng, Portia Opoku Boadi, et al. Estimation of biogas and methane yields in an UASB treating potato starch processing waste water with backpropagation artificial neural network[J]. Bioresource Technology, 2017, 228: 106-115. doi: 10.1016/j.biortech.2016.12.045 [5] YU Xin, LV Ruihong, SONG Fang, et al. Pocket-sized nondispersive infrared methane detection device using two-parameter temperature compensation[J]. Spectroscopy Letters, 2014, 47(1): 30-37. doi: 10.1080/00387010.2013.780082 [6] Reich S L, Gomez D R, Dawidowski L E. Artificial neural network for the identification of unknown air pollution sources[J]. Atmospheric Environment, 1999, 33(18): 3045—3052. doi: 10.1016/S1352-2310(98)00418-X [7] SHAO Jie, TANG Qianjin, CHENG Cheng, et al. Remote detection of alcohol concentration in vehicle based on TDLAS[C]//Symposium on Photonics and Optoelectronic, 2010: 1-3. [8] 耿辉, 张玉钧, 刘文清, 等. 乙醇蒸气近红外高分辨光谱获取方法研究[J]. 大气与环境光报, 2012, 7(1): 57-62.GENG Hui, ZHANG Yujun, LIU Wenqing, et al. Acquisition method of high resolution spectra of ethanol vapor in near-IR range[J]. Journal of Atmospheric and Environmental Optics, 2012, 7(1): 57-62. [9] KAN Ruifeng, LIU Wenqing, ZHANG Yujun, et al. Large scale gas leakage monitoring with tunable diode laser absorption spectroscopy[J]. Chin. Opt. Lett. , 2006, 4(2): 116-118. http://www.cqvip.com/qk/85954X/200602/21084242.html [10] 吴涛, 张怀林, 孔维平, 等. 基于2.73 μm分布反馈式激光器同时在线测量呼出气体中的CO2和水汽[J]. 光学学报, 2018, 38(12): 1230001-8. https://www.cnki.com.cn/Article/CJFDTOTAL-GXXB201812050.htmWU Tao, ZHANG Huailin, KONG Weiping, et al. Simultaneous on-line measurement of exhaled carbon dioxide and water vapor based on 2.73 μm distributed feedback laser[J]. Acta Optica Sinica, 2018, 38(12): 1230001-8. https://www.cnki.com.cn/Article/CJFDTOTAL-GXXB201812050.htm [11] 陈磊, 李长俊. 基于BP神经网络预测硫在高含硫气体中溶解度[J]. 石油与天然气化工, 2015, 44(3): 1-5. doi: 10.3969/j.issn.1007-3426.2015.03.001CHEN Lei, LI Changjun. Prediction of sulfur solubility in high sulfur gas based on BP neural network[J]. Chemical Engineering of Oil & Gas, 2015, 44(3): 1-5. doi: 10.3969/j.issn.1007-3426.2015.03.001 [12] 刘国华. 机动车尾气高温CO、CO2排放因子检测关键技术研究[D]. 北京: 中国科学技术大学, 2018.LIU Guohua. Research on Key Technologies of Vehicle Exhaust Temperature CO and CO2 Emissions Factor Detection[D]. Beijing: University of Science and Technology of China, 2018.