基于支持向量机的煤矿极性气体光谱定量分析

发布时间：2018-08-20 17:25

【摘要】：目前煤矿井下极性气体定量分析和检测方法存在的问题,初步建立了基于支持向量机算法的极性气体定量分析校正模型;利用傅里叶变换红外光谱法对CH4、CO2、SF6等10种煤矿井下极性气体进行扫描,得出共310组检验样本数据,通过光谱数据预处理和SVM校正模型的训练与检验,实现了SVM校正模型,使模型预测精度基本达到煤矿气体的检测要求。经过模型验证,除C3H8、n C4H10正丁烷和CO2的准确度较低外,其他样本测试浓度和实际浓度偏差均不超过该组分满量程的10%,通过对理论值与测量值对比,分析结果满足现场实际需求,初步实现了利用红外光谱技术对煤矿井下极性气体的定量分析,在建模速度和模型预测精度等方面均取得了较好的效果,达到了快速检测分析井下灾害极性气体浓度的目的。
[Abstract]:At present, there are some problems in the quantitative analysis and detection method of polar gas in coal mine, and the correction model of polar gas quantitative analysis based on support vector machine is established preliminarily. By using Fourier transform infrared spectroscopy (FTIR) to scan 10 kinds of polar gases in coal mine, such as Ch _ 4C _ 2CO _ 2H _ 6 and so on, a total of 310 groups of test sample data are obtained. The SVM correction model is realized through the pretreatment of spectral data and the training and testing of SVM correction model. The prediction accuracy of the model is basically up to the requirement of coal mine gas detection. After model verification, except for the low accuracy of C3H8n C4H10 n-butane and CO2, the deviation of test concentration and actual concentration of other samples does not exceed 10% of the full range of the component. By comparing the theoretical values with the measured values, the results of the analysis meet the actual demand in the field. The quantitative analysis of polar gas in coal mine by using infrared spectrum technology has been preliminarily realized. Good results have been achieved in modeling speed and model prediction accuracy, and the purpose of rapid detection and analysis of underground polar gas concentration has been achieved.
【作者单位】：辽宁石油化工大学顺华能源学院;煤科集团沈阳研究院有限公司;煤矿安全技术国家重点实验室;中国矿业大学安全工程学院;
【基金】：国家重大科学仪器设备开发专项资助项目(2012YQ24012705) “十二五”国家科技支撑计划资助项目(2012BAK04B09) 国家重点基础研究发展计划(973计划)资助项目(2011CB201206) 中煤科工集团科技创新基金资助项目(2014MS030)
【分类号】：TD711;O657.3

【参考文献】