樱桃多品质数据分析与无损快速检测模型的建立

Analysis of the multi-quality data and the development of the non-destructive rapid testing models of cherry

中文关键词:  近红外光谱法  无损检测  可溶性固形物含量  可滴定酸含量  果实硬度  樱桃

英文关键词:near infrared spectroscopy  non-destructive testing  soluble solid content  titratable acid content  fruit firmness  cherry

中文摘要:

      目的 研究樱桃多品质数据分布情况, 建立樱桃多品质无损快速检测方法。方法 对樱桃样品分别测试可溶性固形物含量(soluble solid content, SSC)、可滴定酸含量(titratable acid content, TAC)、果实硬度(fruit firmness, Firm)。采用统计分析方法对数据进行统计学描述, 分别绘制含量分布直方图并计算直方图分布频次百分比。以樱桃样品近红外(near infrared, NIR)光谱数据为自变量、品质数据参考值为因变量建立樱桃品质无损快速定量检测模型。结果 统计分析结果表明, 可溶性固形物含量11~17 Brix区间范围内的样品数占样品总数的约86.0%, 可滴定酸含量0.1%~0.8%区间范围内的样品数占样品总数的约90.4%, 果实硬度1.60~ 3.00 kg/cm2区间范围内的样品数占样品总数的约86.0%。多元回归建模结果表明, 剔除异常值有助于提高模型预测性能, 剔除异常值后可溶性固形物含量、可滴定酸含量、果实硬度模型的相对预测性能值分别提高了15.3%、32.9%、12.3%。结论 采用统计分析结合直方图分析可较直观地描述樱桃品质分布情况; 剔除异常值对提高樱桃可滴定酸含量近红外无损检测模型预测能力的作用最大。

英文摘要:

      Objective To study the distribution of the multi-quality data of cherry, and develop the non-destructive rapid testing methods for cherry. Methods The cherry samples were tested for soluble solid content (SSC), titratable acid content (TAC) and fruit firmness (Firm). The statistics methods were applied to describe the statistical characteristics of the data, meanwhile the histogram of the content distribution was drawn with the percentage of frequency in the histograms respectively. The non-destructive rapid quantitative calibration models were developed by the near infrared (NIR) spectra data of cherry as independent and the specified values of the qualities as dependent. Results It was demonstrated by statistic analysis that the percentage of the samples with SSC between 11 and 17 Brix was about 86.0% of the total samples, the percentage of the samples with TAC between 0.1% and 0.8% was about 90.4% and the percentage of the samples with Firm between 1.60-3.00 kg/cm2 was about 86.0%. It was indicated by multi-regression models that the outlier elimination was good for enhancing the prediction performance of the models, by which, the ratio performance deviation values had been increased by 15.3%, 32.9%, 12.3% for the models of SSC, TAC and Firm respectively. Conclusion Statistical analysis combined with histogram analysis can directly describe the distribution of cherry quality, eliminating outliers has the greatest effect on improving the prediction ability of NIR nondestructive testing model of titratable acid content in cherry.