【摘要】 快速分析小麦秸秆中K和Na元素含量对提高其燃烧效率具有重要的现实意义。选用华北地区29个小麦秸秆代表性样本作为研究对象,以电感耦合等离子体质谱法(ICP-MS)量测结果作为标准值,探讨激光诱导击穿光谱(LIBS)技术对小麦秸秆中K和Na元素含量进行定量预测分析的可行性。为提高模型定量分析精度,首先分别选取K和Na分析线附近波段光谱作为定标模型原始光谱数据,对比基线校正(BC)、归一化(Norm)与中心化(MC)相互组合算法对LIBS光谱降噪效果影响,分析比较线性建模方法:偏最小二乘回归(PLSR)和非线性建模方法:增强型反向传播人工神经网络(BP-ADaboost)对预处理后光谱数据的适用性。研究结果发现,与PLSR模型相比较,小麦秸秆中K和Na的BP-ADaboost最优模型效果均较好,其预测决定系数R2p分别为0. 908和0. 979,预测均方根误差分别为2. 388 g/kg和0. 138 g/kg,相对分析误差分别为2. 358和4. 203。结果表明,LIBS技术能用于小麦秸秆中K和Na的同步快速定量分析。更多还原

【Abstract】 It is of great practical significance to rapidly analyze the content of potassium( K) and sodium( Na) in wheat straw for improving its combustion efficiency. And totally 29 representative wheat straw samples collected from North China were chosen as the research objects. Based on the standard values measured by inductively coupled plasma mass spectrometry( ICP-MS), laser induced breakdown spectroscopy( LIBS) was used for the quantitative analysis of K and Na contents in wheat straw. In order to improve the accuracy of quantitative analysis,the spectral bands around the analytical lines of K and Na were primarily confirmed as original spectral data of the calibration models,respectively. The effects of baseline correction( BC),normalization( Norm) and mean-centering( MC) on LIBS spectral denoising were compared. Moreover,the applicability of partial least squares regression( PLSR) and Adaboost backpropagation artificial neural network( BP-ADaboost) for preprocessed spectral data was compared and analyzed. Results showed that when compared with PLSR models,the BP-ADaboost models of potassium and sodium in wheat straw both had better effects,yielding R2 pof 0. 908 and 0. 979,root mean square error of prediction set of 2. 388 g/kg and 0. 138 g/kg,relative percent deviation of2. 358 and 4. 203,respectively. Therefore,LIBS technique can be used for the simultaneous quantitative analysis of K and Na in wheat straw.更多还原