基于介电特性的灵武长枣新鲜度预测  

Prediction on freshness degree of Lingwu long jujube on dielectric properties

作　　者：沈静波[1] 张海红[1] 马雪莲[1] 王慧倩[1] 李子文[1] 周世平[1] 

机构地区：[1]宁夏大学农学院,宁夏银川750021

出　　处：《食品与机械》2016年第1期117-120,212,共5页Food and Machinery

基　　金：国家自然科学基金资助项目(编号:31160346)

摘　　要：为了研究灵武长枣新鲜度与介电特性参数的关系,利用LCR测试仪在1.995kHz下测试长枣的介电特性参数,并对其介电特性参数和品质参数进行相关性分析。结果表明:长枣的介电损耗因子ε″与可溶性固形物含量、硬度、失重率、可滴定酸含量和丙二醛含量呈极显著相关(P<0.01),相对相对介电常数ε′仅与呼吸强度显著相关(P<0.01)。根据可溶性固形物含量、硬度和失重率的变化规律,将长枣分为3个新鲜度等级。以介电损耗因子ε″为BP神经网络的输入特征参数,利用BP神经网络结构建立长枣的新鲜度预测模型,新鲜度等级平均识别率达到81.67%,可用来预测灵武长枣的新鲜度。In order to study the relationship between freshness degree and dielectric properties of Lingwu long jujube, the dielectric properties parameters of long jujube were measured using （LCR）on 1. 995 kHz, and the correlation analysis between dielectric properties parameters and quality parameters was conducted. The results indi cated that correlations between long jujube＇s dielectric loss factor ε＂ and soluble solids content, hardness, weight loss, titratable acids content ＆ MDA content were very significant （P〈0. 01）, and the correlation between relative dielectric constant e＇ and respiration intensity was very significant （P〈0. 01）. The long jujube were divided into three freshness grades according to the change rules of soluble solids content, hardness and weight loss rate. Using dielectric loss factor ε＂ as the input characteristic parameters, the prediction model of long jujube freshness was established by BP neural network structure. The average distinguishing rate of freshness grades was 81. 67%. The results indicated that the freshness degree of long ju jube could be predicted by dielectric properties parameters.

关 键 词：枣 介电特性 新鲜度 BP神经网络 

分 类 号：TS255.7[轻工技术与工程—农产品加工及贮藏工程]