CAJ | 학술논문

　　针对目前面粉灰分含量的检测方法存在操作繁琐、耗时长、费时费力和检测效率低等问题，运用近红外光谱分析技术检测面粉的灰分含量，选择最优的光谱预处理方法和光谱范围，采用偏最小二乘法（ PLS）及 BP 神经网络算法进行定量分析研究。结果表明：采用偏最小二乘法（ PLS）所建的定量分析模型的决定系数R 2为90．66，预测均方根误差RMSEP为0．0553，总偏差为0．02793；用BP神经网络预测总偏差为0．0367。研究发现，近红外光谱技术用于快速无损检测面粉灰分含量是可行的，且PLS、BP 神经网络算法可进行面粉灰分含量预测。
　　침대목전면분회분함량적검측방법존재조작번쇄、모시장、비시비력화검측효솔저등문제，운용근홍외광보분석기술검측면분적회분함량，선택최우적광보예처리방법화광보범위，채용편최소이승법（ PLS）급 BP 신경망락산법진행정량분석연구。결과표명：채용편최소이승법（ PLS）소건적정량분석모형적결정계수R 2위90．66，예측균방근오차RMSEP위0．0553，총편차위0．02793；용BP신경망락예측총편차위0．0367。연구발현，근홍외광보기술용우쾌속무손검측면분회분함량시가행적，차PLS、BP 신경망락산법가진행면분회분함량예측。
Aiming at the ash content in the flour detection methods exist the tedious , time-consuming, operation time-consuming detection efficiency and low , using near-infrared spectroscopy to detect the ash content in the detection of flour,select the optimal pretreatment methods and the spectra of the spectral range , using partial least squares (PLS) and the BP neural network algorithm quantitative analysis .The results showed that the correlation coefficient can be 0.958 1, root mean square error of prediction RMSEP can be 0.055 3,and the total deviation can be 0.059 07 by partial least squares (PLS) quantitative analysis model ,with BP neural network to predict the total deviation can be 0.036 7.The study proved it was feasible to use near-infrared spectroscopy on rapid non-destructive testing of ash content in the flour , PLS and BP neural network algorithm can be used to forecast the content of flour .