In this paper, we present simultaneous multiple pollutant gases (CO2, CO, and NO) measurements by using the non-dispersive infrared (NDIR) technique. A cross-correlation correction method is proposed and used to c...In this paper, we present simultaneous multiple pollutant gases (CO2, CO, and NO) measurements by using the non-dispersive infrared (NDIR) technique. A cross-correlation correction method is proposed and used to correct the cross-interferences among the target gases. The calculation of calibration curves is based on least-square fittings with third-order polynomials, and the interference functions are approximated by linear curves. The pure absorbance of each gas is obtained by solving three simultaneous equations using the fitted interference functions. Through the interference correction, the signal created at each filter channel only depends on the absorption of the intended gas. Gas mixture samples with different concentrations of CO2, CO, and NO are pumped into the sample cell for analysis. The results show that the measurement error of each gas is less than 4.5%.展开更多
为了实现柿子(Diospyros kaki thunb)可溶性固形物含量的快速无损检测,提出了一种采用可见-近红外光谱分析技术无损检测柿子可溶性固形物含量的方法。采用FieldSpec3光谱仪对3种不同品种的柿子进行光谱分析,共获取66个样本数据。利用平...为了实现柿子(Diospyros kaki thunb)可溶性固形物含量的快速无损检测,提出了一种采用可见-近红外光谱分析技术无损检测柿子可溶性固形物含量的方法。采用FieldSpec3光谱仪对3种不同品种的柿子进行光谱分析,共获取66个样本数据。利用平均平滑法对样本数据进行预处理,再采用主成分分析法,依据可信度获取光谱的6个主成分数据。将样本随机分成51个建模样本(每种各17个)和15个验证样本(每种各5个),把6个主成分数据作为BP神经网络的输入变量,柿子的可溶性固形物含量作为输出变量,隐含层的节点数为11,建立3层BP神经网络检测模型,并用该模型对15个验证样本进行预测。结果表明,所建校正模型的校正标准差(SEC)为0.232,对预测集样本可溶性固形物含量的预测相对误差在3%以下,预测值和实测值的决定系数(R2)为0.99,预测标准差(SEP)为0.257。结果表明应用近红外光谱技术结合主成分分析和神经网络算法检测柿子的可溶性固形物含量是可行的。展开更多
基金Project supported by the National High Technology Research and Development Program of China (Grant No. 2009AA063006)the National Natural Science Foundation of China (Grant No. 40805015)the Excellent Youth Scientific Foundation of Anhui Province, China (Grant No. 10040606Y28)
文摘In this paper, we present simultaneous multiple pollutant gases (CO2, CO, and NO) measurements by using the non-dispersive infrared (NDIR) technique. A cross-correlation correction method is proposed and used to correct the cross-interferences among the target gases. The calculation of calibration curves is based on least-square fittings with third-order polynomials, and the interference functions are approximated by linear curves. The pure absorbance of each gas is obtained by solving three simultaneous equations using the fitted interference functions. Through the interference correction, the signal created at each filter channel only depends on the absorption of the intended gas. Gas mixture samples with different concentrations of CO2, CO, and NO are pumped into the sample cell for analysis. The results show that the measurement error of each gas is less than 4.5%.
文摘为了实现柿子(Diospyros kaki thunb)可溶性固形物含量的快速无损检测,提出了一种采用可见-近红外光谱分析技术无损检测柿子可溶性固形物含量的方法。采用FieldSpec3光谱仪对3种不同品种的柿子进行光谱分析,共获取66个样本数据。利用平均平滑法对样本数据进行预处理,再采用主成分分析法,依据可信度获取光谱的6个主成分数据。将样本随机分成51个建模样本(每种各17个)和15个验证样本(每种各5个),把6个主成分数据作为BP神经网络的输入变量,柿子的可溶性固形物含量作为输出变量,隐含层的节点数为11,建立3层BP神经网络检测模型,并用该模型对15个验证样本进行预测。结果表明,所建校正模型的校正标准差(SEC)为0.232,对预测集样本可溶性固形物含量的预测相对误差在3%以下,预测值和实测值的决定系数(R2)为0.99,预测标准差(SEP)为0.257。结果表明应用近红外光谱技术结合主成分分析和神经网络算法检测柿子的可溶性固形物含量是可行的。