基于聚核模糊分类的多指标水蜜桃成熟度判别  

Fuzzy classification with kernel clustering for peach maturity discrimination with multi-dimensional indexes

作　　者：江亿平 卞贝 张兆同[1] 潘磊庆[2] 汪小旵[3] JIANG Yiping;BIAN Bei;ZHANG Zhaotong;PAN Leiqing;WANG Xiaochan(College of Information Management,Nanjing Agricultural University,Nanjing 210095,China;College of Food Science and Technology,Nanjing Agricultural University,Nanjing 210095,China;College of Engineering,Nanjing Agricultural University,Nanjing 210031,China)

机构地区：[1]南京农业大学信息管理学院,江苏南京210095 [2]南京农业大学食品科学技术学院,江苏南京210095 [3]南京农业大学工学院,江苏南京210031

出　　处：《食品与发酵工业》2021年第9期174-182,共9页Food and Fermentation Industries

基　　金：国家自然科学基金资助项目(71803084);教育部人文社会科学基金项目(17YJC630048);中央高校基本科研业务费项目(NAU:SKCX2020009)。

摘　　要：针对水蜜桃成熟度判别的模糊性和不确定性,提出基于聚核模糊分类的多维指标水蜜桃成熟度判别方法。首先,选择与水蜜桃成熟相关的出汁率、糖度、硬度和失重率指标,构建多维指标数据集。其次,根据数据集分布和模糊区域重叠度,建立半梯半岭型隶属度函数。最后,利用熵值法计算多维指标输出权重集,并引入聚核权向量组融合相邻成熟阶段信息,建立基于聚核模糊分类的多维指标水蜜桃成熟度判别模型。结果表明,影响水蜜桃成熟度的多维指标权重由大到小依次为:糖度、出汁率、硬度和失重率;所提模型能较为准确地识别水蜜桃成熟等级,正确率为93.75%;对比常见的三角、梯形隶属度函数和最大隶属度法,聚核模糊分类能够提高判别正确率2.08%~12.50%。该结果为水蜜桃食品加工提供科学可靠的品质划分依据,能够提高食品加工效率、保障桃加工产品品质。Peach maturity discrimination is significant to improve the efficiency and quality of peach products.To discriminate peach maturity,a fuzzy classification with kernel clustering by using multi-dimensional indexes was proposed.Firstly,according to the trend of maturity indexes during peach growing,the juice yield,sugar degree,firmness,and weight-loss ratio indexes were selected to establish a multi-dimensional index data set.Then,the mappings between maturity indexes and stages were interval and fuzzy.The membership functions of semi-ladder and semi-ridge type were formulated by integrating data distribution and overlap degree of fuzzy regions.Finally,the entropy method was used to determine the index output weight set considering dispersions of the membership matrix.A weight index group of kernel clustering was proposed to merge information at adjacent stages,which could improve the discrimination performance of peach maturity.The results revealed that the weight of multi-dimensional indexes affecting peach maturity was Brix value,juice yield,firmness and weight-loss ratio in descending order.The proposed method could distinguish peach maturity stages with an accuracy rate of 93.75%,which realized the effective discrimination of peach four maturity stages.Compared with the traditional triangular,trapezoidal type membership functions and maximum membership principle,the accuracy rates of the proposed method have been increased by 2.08%-12.50%.The proposed method for peach maturity discrimination provides a scientific and accurate quality standard for peach food processing,which could improve food processing efficiency and ensure the quality of peach products.

关 键 词：水蜜桃 成熟度判别 多维指标集 隶属度 聚核模糊分类 

分 类 号：TS255.1[轻工技术与工程—农产品加工及贮藏工程]