Analysis of NIR Characteristic Wavelengths for Apple Flesh Firmness Based on GA and iPLS
TU Zhen-hua1, JI Bao-ping1, MENG Chao-ying2, ZHU Da-zhou1, SHI Bo-lin3, QING Zhao-shen1*
1. College of Food Science and Nutritional Engineering, China Agricultural University, Beijing 100083, China 2. College of Information and Electrical Engineering, China Agricultural University, Beijing 100083, China 3. Institute of Food and Agriculture Standardization, China National Institute of Standardization, Beijing 100088, China
摘要： 利用傅里叶近红外光谱(FT-NIRS)测定了苹果的硬度。通过使用几种基于遗传算法和间隔偏最小二乘法的特征波长选取方法，包括动态向后间隔偏最小二乘(dynamic backward version of interval PLS，dynamic biPLS)、动态向后间隔偏最小二乘结合遗传算法(dynamic biPLS & GA-PLS)和反复的遗传算法(iterative GA-PLS)，分析了苹果硬度的特征波长。结果表明，运用遗传算法和间隔偏最小二乘选择特征波长后，不但可以降低模型的复杂度，同时能够达到提高模型预测精度的效果。在此基础上，研究分析了苹果硬度特征波长的物理化学意义。由于果胶是在苹果成熟过程中一种和硬度有很大关联的物质，通过比较苹果硬度的特征波长和果胶的特征吸收峰，发现两者具有有很好的一致性。因此，采用遗传算法和间隔偏最小二乘法得到的苹果硬度的特征波长能够反映果胶的吸收信息，从而解释了近红外技术检测苹果硬度的机理。
关键词：近红外光谱;硬度;遗传算法;间隔偏最小二乘法;苹果;果胶
Abstract：In the present study, the fruit flesh firmness of apple was analyzed by near infrared (NIR) spectroscopy using an FT-NIR spectrometer. The sensitive spectral regions that provide the lowest prediction error were analyzed by different well-known variable selection methods, including dynamic backward interval partial least-squares (dynamic biPLS), sequential application of backward interval partial least-squares and genetic algorithm(dynamic biPLS & GA-PLS), and iterative genetic algorithm partial least-squares (iterative GA-PLS). Iterative GA-PLS, dynamic biPLS & GA-PLS led to a distinct reduction in the number of spectral data points with better predictive quality. Furthermore, the majority of selected wavelengths were content with the characteristic of the sorption bands of fruit flesh firmness. Pectin constituents, complex non-starch polysaccharides, which are related to texture change in apple, play an important role in their harvest maturity, ripening and storage. Comparing NIR characteristic wavelengths of apple flesh firmness and typical absorption bands for pectin, it was found that characteristic wavelengths of apple flesh firmness were consistent with the pectins relevant spectral regions. Therefore, the NIR characteristic wavelengths of apple firmness based on GA and iPLS reflected the chemical component of apple and the results were reasonable.
屠振华1,籍保平1,孟超英2,朱大洲1,史波林3,庆兆珅1* . 基于遗传算法和间隔偏最小二乘的苹果硬度特征波长分析研究[J]. 光谱学与光谱分析, 2009, 29(10): 2760-2764.
TU Zhen-hua1, JI Bao-ping1, MENG Chao-ying2, ZHU Da-zhou1, SHI Bo-lin3, QING Zhao-shen1* . Analysis of NIR Characteristic Wavelengths for Apple Flesh Firmness Based on GA and iPLS . SPECTROSCOPY AND SPECTRAL ANALYSIS, 2009, 29(10): 2760-2764.
[1] Abbott J A. Postharvest Biology and Technology, 1999, 15: 207. [2] JIAO Qun-ying, WANG Shu-mao(焦群英, 王书茂). Advances in Mechanics(力学进展), 1999, 29(4): 583. [3] Dull G, Birth G S, Smittle D A, et al. Journal of Food Science, 1989, 54(2): 393. [4] Lu R, Ariana D. Applied Engineering in Agriculture, 2002, 18: 585. [5] Gómez H A, He Y, Pereira A G. Journal of Food Engineering, 2006, 77: 313. [6] Leardi R. Journal of Chemometrics, 1994, 8(1): 65. [7] Leardi R, González A L. Chemometrics and Intelligent Laboratory Systems, 1998, 41: 195. [8] Norgaard L, Saudland A, Wagner J, et al. Applied Spectroscopy, 2000, 54: 413. [9] Kennard R W, Stone L A. Technometrics, 1969, 11: 137. [10] Knee M, Bartley I M. Composition and Metabolism of Cell Wall Polysaccharides in Ripening Apples. In Friend J & Rhodes M J C(Eds.), Recent Advances in the Biochemistry of Fruit and Vegetables. New York: Academic Press, 1981. 133. [11] Van Buren J P. Function of Pectin in Plant Tissue Structure and Firmness. In Walter R H(Ed.), The Chemistry and Technology of Pectin. New York: Academic Press, 1991. 1. [12] Boeriu C G, Stolle-Smits T, Van Dijk C. Journal of Near Infrared Spectroscopy, 1998, 6: A299. [13] Sohn M R, Cho R K. Journal of Korean Society for Horticultural Science, 2000, 41: 65. [14] Sirisomboon P, Tanaka M, Fujita S, et al. Journal of Food Engineering, 2007, 78: 701.
