论文部分内容阅读
自行搭建了气体采集系统,根据井下的气体情况,采集了包括甲烷、乙烷、丙烷、正丁烷和二氧化碳五种气体的中红外光谱数据共236组,其中校正集186组,验证集50组。在对光谱数据进行预处理之后,利用主成分分析技术将得到的主吸收峰区域的红外光谱数据进行降维处理,通过特征提取得到3个特征值作为矿井气体光谱数据的输入量。该方法有效减少了模型的计算量,加快了模型的收敛速度。然后,利用改进支持向量机分别对这五种气体建立了定量分析模型。为提高该算法的预测精度,利用遗传算法和粒子群优化算法分别对SVM参数进行参数寻优。最后,选择优化效果更好的粒子群算法,并通过验证集对这五种气体进行了浓度预测分析。实验结果表明:五种气体浓度预测结果的平均误差均小于1.78%,最大误差均小于4.98%,且对于50组的气体预测耗时均小于103 s。表明所提出的改进的SVM算法能够准确、快速地预测矿井气体浓度,对实现矿井气体检测有着积极的意义。
A total of 236 mid-infrared spectral data including methane, ethane, propane, n-butane and carbon dioxide gas were collected based on the gas conditions in the well, of which 186 calibration sets, 50 validation sets . After preprocessing the spectral data, the principal component analysis (PCA) technique was used to reduce the infrared spectral data of the main absorption region, and the three eigenvalues were extracted by the feature extraction as the input of spectral data of mine gas. The method effectively reduces the computational complexity of the model and accelerates the convergence rate of the model. Then, a quantitative analysis model of these five kinds of gases is established by using improved SVM. In order to improve the prediction accuracy of the algorithm, genetic algorithm and particle swarm optimization algorithm are used to optimize the parameters of SVM respectively. Finally, the particle swarm optimization with better optimization is selected, and the concentrations of these five gases are predicted by the verification set. The experimental results show that the average errors of the prediction results of the five gas concentrations are less than 1.78% and the maximum errors are all less than 4.98%, and the gas prediction time for the 50 groups is less than 103 s. The results show that the proposed SVM algorithm can accurately and quickly predict the gas concentration in coal mines and has a positive effect on gas detection in mines.