基于LIBS和遗传算法的畜禽粪便中钙含量研究

为实现畜禽粪便高效、安全资源化利用，快速、准确定量表征分析畜禽粪便中钙(Ca)元素含量具有重要意义。以我国畜禽养殖比较集中地区的代表性畜禽粪便样本为对象，探讨了LIBS快速、定量分析其Ca元素含量模型构建和通过GA筛选变量优化模型的可行性。采集光谱的仪器参数设置如下：采集点数、激光能量、压制样压力、光斑大小和延迟时间分别为80个，15%，20 T，400 μm和1 μs。建模结果表明：采用基于Ca元素特征波长所建三种线性模型精度较低；基于全谱的PLS所建模型中，R2_v和RPD分别为0.85和2.13；选用化学计量学方法-遗传算法进行变量筛选可显著减少建模变量信息，提高建模效率，经GA优选所得12个变量所建PLS模型中R2_v和RPD为0.90和3.04，具有较高精度和实际应用价值。研究表明，将激光诱导击穿光谱技术和遗传算法相结合可定量分析畜禽粪便中Ca元素含量。

Modelling of Calcuim Content in Manure Using Laser-Induced Breakdown Spectroscopy and Genetic Algorithm Combined with Partial Least Squares

In order to enhance the efficiency and safety of manure resource utilization,a rapid quantitative analysis of calcuim (Ca) content in manure is of great significance.In the presented study,the application of laser-induced breakdown spectroscopy (LIBS) technique was used to quantitatively analyze Ca content in manure.Genetic algorithm (GA) was also applied to the LIBS to optimize the model.The dominant factors of LIBS were set: 80 collection dots with 15% laser energy,400 μm spot size,delay time of 1.0 μs,and preforming pressure at 20 tons (T).The modeling results showed that the initial linear model constructed from the characteristic wavebands of Ca presented low precision and accuracy,partial least squares (PLS) models with wavebands at 190~950 nm exhibited the effects with coefficient of determination for the validation set (R2v) and relative prediction deviation (RPD) of 0.85 and 2.13,respectively.The PLS model considered 12 variables,which were selected with GA in the waveband at 190~950 nm,and presented the R2v and RPD of 0.90 and 3.04,respectively.They had presented relatively better results by avoiding complicated sample processing.It is pertinent to note that the efficiency of this method increased by a large margin when the variables were selected based on GA analysis.The results showed that LIBS combined with GA can be used for quantitative analysis of Ca in manure.