论文部分内容阅读
为快速、便捷地利用近红外光谱检测苹果糖度,设计了以ARM9处理器为核心、以微型光谱仪和自制果托作为光谱检测装置、以Win CE为操作系统的便携式苹果糖度光谱检测仪。以80个苹果样品作为试验对象,采用平滑、多元散射校正、标准正态变量变换等方法对原始光谱进行预处理,结合无信息变量消除法和连续投影算法进行有效波长的筛选,建立基于所选特征波长和全波段的苹果糖度近红外光谱偏最小二乘模型。结果表明,偏最小二乘结合原始光谱信息建模效果最好,其预测相关系数Rp=0.853,预测均方根误差RMSEP=0.534。该检测仪能较好地满足苹果糖度的快速无损检测。该研究为快速、便携的苹果糖度光谱检测仪设计提供了参考。
In order to detect the apple sugar content quickly and conveniently by near infrared spectroscopy, a portable apple sugar degree spectrum detector with ARM9 processor as its core, micro spectrometer and homemade fruit holder as spectral detection device and Win CE as operating system was designed. Eighty apple samples were selected as experimental subjects. The original spectra were preprocessed by using smoothing, multivariate scatter correction and standard normal transformation, and the effective wavelength was filtered by the combination of no information variable eliminating method and continuous projection algorithm. Partial least square model of apple sugar content near infrared spectra with characteristic wavelength and full band. The results show that partial least squares combined with the original spectral information modeling is the best, the correlation coefficient of prediction Rp = 0.853, the root mean square error of prediction RMSEP = 0.534. The detector can better meet the fastness non-destructive detection of apple sugar. This study provides a reference for the design of fast and portable apple sugar spectrum detector.