西瓜光透射规律与品质属性的内在联系

Interent Relation Between the Light Transmission Law and Quality Attributes of Watermelon

果品内部化学基础信息与光谱信息较好对应是提高模型的关键。大量的皮厚、体积大的水果果肉在可见-近红外区域透光性差、光折射角度难以确定、化学基础信息获取不准,导致品质预测效果差。以西瓜为研究对象,对西瓜不同区域的可溶性固形物与品质属性的内在联系进行探讨。水果市场购买360个西瓜样品,在线检测装置采集光谱时参数用两种设置:积分时间100 ms、电流8.0 A和积分时间为150 ms,电流8.15 A,后者的光谱吸收峰值强度更高。西瓜可溶性固形物含量测定时,将西瓜分为8份,分别测量心糖、中糖、外围糖、底边糖(SSC)和混合糖(SSC)的平均值,西瓜内部不同区域可溶性固形物有较大的差异,果中心的心糖值最高,而越靠近瓜皮区域的糖度值越低。以西瓜不同区域可溶性固形物为因变量,卷积平滑(S-G)降解光谱噪声后的光谱为自变量建立可溶性固形物偏最小二乘预测模型,建模集270个,预测集90个。对比模型发现,提高分选装置的积分时间和卤钨灯电流,可以增加可溶性固形物模型预测精度;局部区域的可溶性固形物作为模型的因变量预测效果也高于混合糖为因变量建立的模型。由于可见-近红外入射后在瓜果内部发生一定角度的折射、光停留在浅层区透射等原因,靠近瓜皮的底边糖区域表征了较多的西瓜果肉信息,建模效果最佳,预测集相关系数为0.89,均方根误差为0.24,建模集相关系数为0.96,均方根误差为0.18。而中糖、外围糖等具有一定深度且在光的直线区域表征的西瓜果实内部信息较少,建模效果较差。因此西瓜底边糖为最佳的可溶性固形物采集区域。研究结果揭示了水果光散射规律特征及其与品质属性的内在联系,可供实现光谱数据库和分析模型的在线更新参数。

Corresponding to the internal chemical basic information of the fruit and the spectral information is the key to improve the model.For the current large thickness and a large volume of fruit,the visible/near infrared is poor in light transmission in the flesh region,the light refraction angle is difficult to determine,and the basic chemical information is not Accurate,resulting in poor quality prediction.In this paper,watermelon is the research object,and the intrinsic relationship between soluble solids and quality attributes in different regions of watermelon is discussed.Three hundred sixty samples are purchased in the fruit market.The parameters of the online detection device are divided into old parameters:integration time 100 ms,current 8.0 A and new parameters.The integration time is 150 ms,the current is 8.15 A to collect the watermelon spectrum,and the spectral absorption peak intensity is higher under the new parameters of the device.When the content of soluble solids in watermelon was determined,the watermelon was divided into 8 parts,and the heart sugar,medium sugar,peripheral sugar,base sugar(SSC)and mixed sugar(SSC)were determined respectively.The soluble solids in different parts of the pulp were large.The difference is that the most recent heart sugar value is the closest to the fruit center,and the lower the sugar value is,the closer it is to the melon skin area.Taking soluble solids in different regions of watermelon as the dependent variable,the new and old parameters of convolution smoothing(S-G)degraded spectral noise was used to establish a soluble solids partial least squares prediction model with independent variables,with 270 modeling sets and 90 prediction sets.The comparison model found that the new parameters improved the integration time of the sorting device and the tungsten halogen lamp current,which increased the prediction accuracy of the soluble solid model;while the soluble solids in the local area as the model’s dependent variable prediction effect was also higher than the mixed sugar.The model established by the variable,the bottom sugar region of the region close to the melon skin,due to visible/near-infrared incidence,and then a certain angle of refraction inside the melon,which indicates more watermelon fruit information,the modeling effect is the best,the prediction set the correlation coefficient is 0.89,and the root mean square error is 0.24.However,the internal information of watermelon fruit characterized by medium sugar and peripheral sugar region is less,and the modeling effect is poor.Therefore,the bottom edge of the watermelon is the best soluble solids collection area.This study reveals the characteristics of fruit light scattering and its intrinsic relationship with quality attributes,and realizes online updating of spectral database and analytical model.