Journal of Nuclear Agricultural Sciences ›› 2021, Vol. 35 ›› Issue (6): 1367-1375.DOI: 10.11869/j.issn.100-8551.2021.06.1367

• Food Irradiation·Food Science • Previous Articles     Next Articles

Quality Petection of Hayward Kiwifruit Based on the Dielectric Properties

LIU Zhenrong, ZHAO Wuqi*, LU Dan, LU Yan, GAO Guitian, MENG Yonghong   

  1. College of Food Engineering and Nutritional Science, Shaanxi Normal University, Xi'an, Shaanxi 710119
  • Received:2020-05-13 Online:2021-06-10 Published:2021-04-16


刘振蓉, 赵武奇*, 卢丹, 路晏, 高贵田, 孟永宏   

  1. 陕西师范大学食品工程与营养科学学院,陕西 西安 710119
  • 通讯作者: *赵武奇,男,副教授,主要从事食品加工新技术研究。
  • 作者简介:刘振蓉,女,主要从事粮食、油脂及植物蛋白质工程研究。
  • 基金资助:

Abstract: To study the mathematical relationship between the dielectric properties(DP) of kiwifruit and its main qualities. The DP, nutrient composition, color difference and texture of hayward kiwifruit in different storage periods were tested. The characteristic frequency points of quality indicators (Vc, soluble solid, ΔE, hardness, cohesiveness and springiness) were selected out by successive projections algorithm(SPA) and genetic algorithm in combination with partial least squares(GA-PLS). The prediction models of kiwi fruit were established based on the dielectric parameters at the characteristic frequency points. The results showed that the optimal models of Vc content and ΔE value was SPA-NN, hardness, soluble solids content, cohesiveness and elasticity was GA-PLS-NN. The determined coefficients of the models were 0.971, 0.934, 0.922, 0.984, 0.908 and 0.954, respectively. There was no significant difference between the predicted value and the measured value(P>0.05) for each quality indicator. The dielectric detection technology combined with SPA and GA-PLS can reflect the characteristic quality of kiwifruit, and the research lays a foundation for the non-destructive detection of kiwifruit property.

Key words: hayward kiwifruit, dielectric properties, quality detection, successive projections algorithm, genetic algorithm-partial least squares(GA-PLS)

摘要: 为研究猕猴桃的介电特性与其主要品质之间的关系,本试验通过测定不同贮藏时期海沃德猕猴桃的介电特性、营养成分、色差及质构指标,用连续投影算法(SPA)和遗传算法-偏最小二乘法(GA-PLS)筛选出各品质指标(Vc、可溶性固形物、ΔE、硬度、粘聚性和弹性)的特征频率点,建立猕猴桃品质指标的预测模型并进行验证。结果表明,建立 Vc和ΔE值的最优模型为连续投影算法结合神经网络模型,可溶性固形物、硬度、粘聚性和弹性的最优模型为遗传算法-偏最小二乘法结合神经网络模型,模型对各指标预测的决定系数分别为0.971、0.934、0.922、0.984、0.908和0.954,各品质指标预测值与实测值无显著差异(P>0.05)。综上所述,介电检测技术结合SPA和GA-PLS可用于预测猕猴桃的特征品质。本研究结果为猕猴桃品质的无损检测奠定了基础。

关键词: 海沃德猕猴桃, 介电特性, 品质检测, 连续投影算法, 遗传算法-偏最小二乘回归(GA-PLS)