灵武长枣采后光谱特征变化的生物光学机理研究

'Lingwuchangzao' jujube is the treasure of Ningxia special fresh jujube products with its rich nutrition and good quality. Since the bio-optical mechanism is still unclear on the spec-tral signature changes during different storage periods of Lingwuchang-zao' jujube. This project intends to take critical internal quality ingredients(such as glucose, malic acid and vitamin C) in 'Lingwuchangzao' jujube as the research object, collect imaging and spectral information of jujube and extract characteristic wavelengths by hyperspectral imaging technology. Then the forecast model of critical internal ingredients in 'Lingwuchangzao' jujube was established. At the same time , jujube samples were irradiated by laser irradiation and scattered fluorescent image was acquired by laser-induced fluorescence hyperspectral imaging technology. Then the fluorescent regions of interest were selected. Characteristic fluorescence peaks were selected in the wavelength range of 400-1700nm.At last, the electronic transition energy corresponding to the characteristic fluorescence peaks was calculated. Combined with characteristic absorption peak in infrared spectroscopy and molecular structure map and quantum mechanics and quantum chemistry theory, the corresponding relationship is built between molecular vibration and atomic energy transition. The theoretical explanation is completed on changes of fluorescence hyperspectral spectral signature of critical internal components in postharvest Lingwuchangzao' jujube, so as to reveal its bio-optical mechanism and provide theoretical reference for quality control of postharvest 'Lingwuchangzao' jujubet during the process of storage.

灵武长枣是宁夏特色鲜食枣珍品，营养丰富、品质优良。针对灵武长枣采后不同贮藏期光谱特征变化的生物光学机理尚不清楚的现状，本项目拟选取灵武长枣关键性的内部品质成分（如葡萄糖、苹果酸、维生素C）为研究对象，采用高光谱成像技术，采集长枣的图谱信息并提取特征波长，建立灵武长枣关键成分的预测模型；同时，采用激光诱导荧光高光谱成像技术，将激光照射到长枣样品后采集诱导出的荧光散射图像，在波长400-1700 nm范围内选取感兴趣荧光区域的荧光特征峰，计算特征荧光峰对应电子跃迁能极差。结合红外光谱特征吸收峰与分子结构谱图、量子力学和量子化学理论，从分子振动和电子能量跃迁的层面建立对应关系，完善灵武长枣采后关键性内部成分的荧光高光谱特征变化的理论解释，从而揭示其背后的生物光学机理，为灵武长枣采后贮藏过程中的品质调控提供理论参考。

利用高光谱及激光诱导荧光高光谱成像技术进行灵武长枣采后关键性的内部成分检测，研究选择有效的特征波长，荧光特征峰、提取特征图像的数学计量学方法，建立灵武长枣采后关键性的内部品质成分的荧光高光谱特征变化的完善理论解释，揭示其光谱特征变化的生物光学机理。在可见-近红外高光谱模型中，IRF-LS-SVM为长枣蔗糖含量的最优预测模型；GAPLS-LS-SVM为长枣果糖含量的最优预测模型；SPA-LS-SVM，为长枣葡萄糖含量的最优预测模型。在近红外高光谱模型中，iPLS-PLSR模型为维生素C最优预测模型；全波段-PLSR模型为草酸和苹果酸最优预测模型；iVISSA-SVM模型为蔗糖最优预测模型，CARS-PLSR模型为果糖和葡萄糖最优预测模型。在激光诱导荧光高光谱系统中，最终确定蔗糖的最优模型为BiPLS-LS-SVM，对应有8条特征波长，分布在737.5~771.1nm之间；果糖的最优模型为CARS-LS-SVM，对应与果糖的相关性较强的波长有4条，分别为660.6、665.4、703.8和756.7nm；葡萄糖的最优模型为GAPLS-LS-SVM，对应关键波长有7条，分别为660.6、665.4、670.2、675.0、775.9、838.3和847.9nm。维生素C含量的预测模型中，SVM模型的预测效果最佳，预测准确率达84.01%；草酸含量的预测模型PLS最优，预测准确率86.2%；苹果酸含量的PLS预测模型最好，预测准确率为84.29%。项目资助发表论文9篇，其中SCI2篇，EI3篇，核心4篇，待发表英文SCI1篇。培养硕士生3名，其中2名已获得硕士学位，1名在读。项目投入经费44万元，累计支出20.7622万元，各项支出与预算基本相符。剩余经费23.2378万元，剩余经费计划用于本项目研究后续支出。