基于高光谱技术的厌氧发酵液固形物含量检测的研究

发布时间：2018-07-06 06:49

本文选题：近红外高光谱 + 总固形物含量　；参考：《光谱学与光谱分析》2017年12期

【摘要】：实时监测发酵液中固形物含量的变化,对控制厌氧发酵过程的稳定性具有重要作用。研究中采用近红外高光谱技术结合化学计量学方法,对水葫芦和稻草秸秆混合厌氧发酵过程中的固形物含量进行定量检测研究。与传统2540G(APHA,1990)标准方法相比,近红外高光谱技术具有无损、快速的优点。实验过程中,首先获取发酵液样本的高光谱信息,应用移动平均平滑法(MAS)进行光谱预处理,并采用竞争自适应重加权采样算法(CARS)、连续投影算法(SPA)和Random frog算法提取光谱特征信息,然后基于全谱和所选特征波长下的光谱信息分别建立总固形物含量(TS)和挥发性固形物含量(VS)的校正模型,建模方法包括偏最小二乘回归(PLSR)和最小二乘-支持向量机(LS-SVM)。研究表明,SPA-LS-SVM模型的预测结果最好,其中TS的预测均方根误差(RMSEP)及相关系数(R_p)分别为0.005 8和0.841;而VS的RMSEP和R_p分别为0.004 1和0.874。结果表明,利用近红外高光谱结合化学计量学方法可以实现厌氧发酵液中的固形物含量的检测,为布置光谱传感器以便定量检测厌氧发酵过程中的固形物含量奠定了理论依据。
[Abstract]:It is important to monitor the change of solid content in fermentation broth in real time to control the stability of anaerobic fermentation process. In this study, the content of solids in the mixed anaerobic fermentation of water hyacinth and straw was studied by using near infrared hyperspectral technique and chemometrics. Compared with the traditional 2540G (APHA 1990) standard method, the near infrared hyperspectral technique has the advantages of nondestructive and fast. In the process of experiment, the hyperspectral information of fermentation broth samples was first obtained, and the spectral characteristics were extracted by moving average smoothing (MAS) method and competitive adaptive re-weighted sampling algorithm (cars), continuous projection algorithm (spa) and Random frog algorithm. Then the calibration models of total solid content (TS) and volatile solid content (vs) are established based on the spectral information under the full spectrum and the selected characteristic wavelengths. The modeling methods include partial least square regression (PLSR) and least squares support vector machine (LS-SVM). The results show that the prediction results of SPA-LS-SVM model are the best, in which the root-mean-square error (RMSEP) and correlation coefficient (RSP) of TS are 0.005 8 and 0.841 respectively, while the RMSEP and RP of vs are 0.004 1 and 0.874 respectively. The results show that the determination of solid content in anaerobic fermentation broth can be realized by using near infrared hyperspectral method combined with chemometrics, which provides a theoretical basis for the arrangement of spectrum sensors for quantitative determination of solid content in anaerobic fermentation process.
【作者单位】：浙江大学生物系统工程与食品科学学院;
【基金】：国家自然科学基金项目(61201073)资助
【分类号】：O657.3;TQ920

【相似文献】