【作者】 周妍；

【导师】 孔晓玲；

【作者基本信息】 安徽农业大学 ， 农业机械化， 2008， 硕士

【摘要】 面条作为我国传统的主食面制品,其品质评价方法至今仍没有建立一个严格、统一的标准。为了适应现代检测技术和食品安全与质量监测的需要,通过简单易行的现场检测,实现综合客观的品质评价,开发新一代便携式食品力学性能检测仪并建立一种客观准确的评价方法势在必行。本文在面条力学性能指标的数据采集、处理与分析的基础上,提出了一种简单易行、客观准确的面条力学性能评价方法。本文在充分调研了国内外对面条力学性能检测状况的基础上,分析和研究了面条品质评价方法的可行性,设计了面条力学性能指标检测的实验方法,并针对46种面条样本做了大量的试验,采集了许多有价值的数据,为面条力学性能的评价起到了积极的作用。本文在原有食品力学性能检测仪的基础上,根据面条的特性研制了检测仪的夹持部件,解决了夹持不稳定的问题,从而可以准确的测量拉力和剪切力,确保了原始数据采集的质量。该仪器可以实现样本的数据属性检测和数据采集。为了对采集的数据进行分析,本文开发了基于数据挖掘技术的食品力学性能分析平台,该平台可以完成样本数据的储存和分析,具有界面友好、简洁直观及可视化等优点,用户可以方便的利用该软件对食品力学性能进行评价。在面条力学性能的评价方法上,本文采用了数据挖掘技术。聚类方法是数据挖掘中的一种重要方法,k-means算法是一种重要的基于划分的聚类分析方法,该方法具有简单、快速、高效及可伸缩性强等优点。本文利用食品力学性能检测仪测量了46种面条样本的各项力学性能指标,将之与面条的其他属性(横截面尺寸等)及环境参数(温度、糊化时间等)组成面条样本的数据集,用k-means算法进行了聚类分析,聚类的结果得到了5个簇,通过对各簇的评价分析发现,每个簇都具有代表该簇的属性特征,这些簇是原数据集的一个有意义的划分,可以用于面条品质的评价。由于聚类分析方法只能进行面条类别的划分,不能对面条品质进行定性描述,所以本文结合主成分分析法(PCA)和感官评价方法,对聚类分析的结果进行了对比分析。将面条品质感官评价和PCA法的结果作为参照,对聚类分析的结果所获得的5个簇进行了评价,得到了面条品质的等级划分及各等级面条品质的描述。本文获得的面条品质评价结果等级划分清晰、用户可通过该实验平台直观地了解待测样本的品质等级、韧性、筋道感、弹性、食味等的好坏。从而解决了面条现场检测的难题。本文所研究的面条品质评价方法,反映了面条品质中与力学特性有关的食品特性,其结果具有较高的灵敏性与客观性,并可对结果进行准确的数量化处理,以量化的指标来客观全面地评价面条品质,从而避免了人为因素对食品品质评价结果的主观影响。更多还原

【Abstract】 As a traditional main food, a strict, uniform method of noodles’ quality evaluation has not established. Aiming at adapting the need of modern examination techniques, food safety and quality inspection, and the target of achieving simple, easy locale examination and integration, objective quality evaluation, it is imperative under the situation that developing a new generation convenient food mechanics performance examination instrument, and establishing an objective evaluation method. Based on the data collection, store and processing of noodles mechanics performance indexes, this dissertation proposes an easy and objective method of noodles mechanics performance evaluation.Based on plenty of researches on the noodles mechanics performance examination status home and abroad, this dissertation analysis and research the possibility of noodles quality evaluation method, and design a experimental method of noodles mechanics performance examinationo We collect plenty of valuable data about the 46 noodles samples through the experiment, which operate a positive effect on noodles mechanics performance evaluation.Based on the original instrument, this dissertation designs the clamper of instrument according to the characters of noodles, which gives a solution of the problem of clamper unstable. This makes the measurement of pull and cut more exactly. This new instrument makes the data indexes examination and data collection come true. For better analysis of the dataset, this dissertation develops a food mechanics performance analysis system based on data mining. This system can achieve the store and analysis of data and it has several characters such as easy use and display the result visually. Users can use this system conveniently to evaluate the mechanics performance of food.This dissertation introduces the data mining technology to the evaluation method of noodles mechanics performance. Clustering analysis is an important method in data mining. K-means algorithm is an important clustering method based on partition. It is simple, fast, high efficiency better retractility. The dataset of noodles samples is compose of each mechanics performance indexes and other attributes. The result of k-means algorithm on this dataset is 5 clusters. Through to evaluation of these clusters, each cluster has the attribute character which can delegate itself. This result is a significative partition of the dataset, which can use to the evaluation of noodles quality. Clustering analysis only can give a qualitative description of noodles quality but only a partition of the noodles. This dissertation compares the clustering result with the PCA and sense evaluation which is considered as a consultant, and than evaluate the S clusters of clustering analysis. The result is the rank about noodles quality and the description of each rank. Through the experiment in this dissertation, the rank of noodles samples is clear, and user can get the intuitionistic information of quality rank, toughness, chewiness, flexibility and taste. And these solve the problem of noodles locale examination.The noodles quality evaluation in this dissertation reflects the food character about the mechanics character in noodles quality. The result is high sensitive and objective. This method can evaluate the noodles quality objectively by the quantitative processing, which avoid the subjective influence by man-made factors in evaluation.更多还原