矿井CO浓度的支持向量机数据融合检测方法

针对检测矿井一氧化碳（CO）含量时，电化学传感器输出受到矿井大气中甲烷气体影响的问题，提出了一种基于支持向量机（SVM）数据融合的CO浓度检测方法。该方法将催化传感器与电化学传感器构成传感器对，利用能够从全局意义上逼近任意非线性关系的支持向量机对传感器对的输出信号进行非线性数据融合，构建了矿井一氧化碳浓度检测模型。实验结果表明，该方法的平均绝对百分比误差为0.88%，均方根误差为1.32 ppm，有效地消除了甲烷对CO电化学传感器的影响，实现了矿井CO浓度的精确检测。

CO concentration detection in coal-mine using Support Vector Machine and data fusion

In order to eliminate the effect of methane gas on electrochemical sensor in coalmine,this paper puts forward a new method for CO concentration detection based on SVM(Support Vector Machine) and data fusion.The method uses SVM,which can approximate nonlinear function in the whole region,to fuse data of sensor pair composed of a catalysis sensor and an elec- trochemical CO sensor,and to get a mathematical model of data fusion for CO concentration detection.The experimental result shows that MAPE(Mean Absolute Percentage Error) is 0.88% and RMSE(Root Mean Square Error) is 1.32%,which indicates the proposed method eliminates effectively the influence of methane gas on electrochemical CO sensor,and realizes accurate CO con- centration detection in coalmine.