【作者】 孙钟雷；

【导师】 孙永海；

【作者基本信息】 吉林大学 ， 农业电气化与自动化， 2012， 博士

【摘要】 食品质地评价是食品品质检验的一个重要方面，它对食品的研究开发、贮藏保鲜、商贸运输等方面都有重要的意义，对人们饮食健康、营养膳食等方面也有一定的指导作用。目前，对食品质地的评价主要采用感官评价法和仪器评价法，感官评价法的评价过程费时费力，评价结果主观性强、而且不稳定；常用的仪器评价法多属于半经验或模拟测定，评价结果与人类感知相差较大，而且不能对多种质地特性做出准确表达。随着食品工业的快速发展和人们对食品品质要求的日益提高，食品质地评价显得越来越重要，市场上急需快速、客观、准确、类人的食品质地评价方法。本文主要针对食品质地评价中存在的问题和评价方法的发展趋势，利用仿生技术、逆向工程、有限元分析等先进技术和理论研究开发咀嚼模拟系统，并使用该系统在多种评价模式下对食品的多个质地指标进行分析检测。1．设计开发了咀嚼模拟系统的硬件装置和测试软件。使用三维激光扫描仪和CT断层扫描仪等设备对人类咀嚼器官模型进行扫描，获取咀嚼系统的形态数据，利用Imageware12、Unigraphics6.0、SolidWorks2008、ANSYS10.0等软件进行咀嚼模拟系统结构设计和优化，然后使用CAD/CAM系统加工制作，主要设计制作了仿生牙齿、仿生牙周膜、仿生颌骨、仿生颞下颌关节、驱动机构、温湿环境部件、辅助机构、信号调理电路等。利用Visual C++6.0软件开发了咀嚼模拟系统测试程序，主要包括信号采集、数据处理、图形分析等模块。对咀嚼模拟系统进行了运动分析和咀嚼功能测试，结果表明咀嚼模拟系统可实现开闭、前后、侧向三维咀嚼运动，与人类咀嚼运动相似；系统的咬合力达到了咀嚼食物的力量，可以咀嚼破碎不同质地的食物；系统的咀嚼效率达到90%以上，和人类无显著差异，且稳定可靠。2．确定了咀嚼模拟系统的测试条件。以9种具有质地代表性的食物样品为试验材料，使用咀嚼模拟系统进行测试，分析了仿生牙周膜、仿咀嚼肌弹簧、电动机转速和温湿环境对咀嚼测试信号的影响，优选出最佳的食品质地测试条件。仿生牙周膜筛选结果表明：仿生下颌第一磨牙底部和周围的仿生牙周膜信号强、重复性好、分类程度高，可作为获取咀嚼模拟系统测试信号的传感器。对仿咀嚼肌弹簧影响分析表明，最佳的弹性系数范围为：测试胡萝卜时，仿咀嚼肌弹簧的弹性系数不小于30N/mm；测试面包时，弹性系数不小于5N/mm；测试糯玉米、豆干、火腿肠、饼干时，弹性系数不小于10N/mm；测试苹果、榨菜、花生时，弹性系数不小于20N/mm。电动机转速对测试信号有一定影响，在对食物质地测试时应保持一个恒定转速，在测试胡萝卜、花生、苹果、糯玉米时，转速范围取60～90r/min；在测试榨菜、豆干、火腿肠、面包、饼干时，转速范围取60～120r/min。温湿环境条件对测试信号影响显著，在对食物进行质地测试时，应该具有适宜温湿环境才能更接近人类的评价，温湿环境为温度37℃、人造唾液流量3mL/min。3．在质地剖面分析（TPA）模式下，对多种食品的多个质地特性进行了分析评价。对胡萝卜、苹果、榨菜、花生、糯玉米、豆干、火腿肠、面包、饼干共9种样品，分别使用咀嚼模拟系统、人类感官、通用食品质构仪进行了硬度、脆性、黏着性、内聚性、弹性、咀嚼性、回复性等质地特性测试。以人类感官评价为基准，通过相关性、多元回归、BP神经网络等分析，建立咀嚼模拟系统的质地剖面分析方法，并与通用食品质地测试仪器进行对比。皮尔逊（Pearson）相关性分析结果表明，各个样品的咀嚼模拟系统测试结果和感官评价结果相关性良好，可以使用咀嚼模拟系统测试代替人类感官评价；咀嚼模拟系统测试结果和感官评价结果的相关性系数均高于通用食品质构仪的测试结果，说明咀嚼模拟系统优于通用食品质构仪。多元回归分析表明，对各个样品建立的质地评价回归模型有效，模型测试和实验测试结果相差小，平均误差和标准残差都较小，差异性结果检验不显著，回归模型满足精度要求。BP神经网络分析表明，对各个样品建立的BP神经网络模型测试效果好，能够满足食品质地评价的要求。4．建立了基于脆裂力学模型的食品脆性评价方法。在咀嚼模拟系统的“杵-臼”咀嚼结构的基础上，通过食品物料在咀嚼破碎过程中的受力分析，建立了脆裂力学模型，并对物料块形状、仿生磨牙形态、咀嚼速度对脆裂力学模型的影响进行了分析，选取胡萝卜、苹果、榨菜三种脆性材料进行了试验验证。结果表明：脆裂方程能正确反映出脆裂力和失效应力之间的关系，可以利用咀嚼模拟系统测得的脆裂力来反映物料的断裂失效情形，可以用来反映物料的脆性。以胡萝卜、苹果、榨菜为试验材料，使用咀嚼模拟系统测试的脆裂信号评价了食物的脆性，并使用感官评价方法进行验证，三点梁弯曲试验进行对比。结果表明：咀嚼模拟系统、三点弯曲试验的测试结果与脆性感官评价都达到了极显著的相关性；咀嚼模拟系统的测试结果与脆性感官评价的相关系数均大于三点弯曲试验；基于咀嚼脆裂信号和感官评分建立的回归方程，满足精度要求，可以测试硬脆性果蔬的脆性。5．在多次咀嚼模式下，利用咀嚼模拟系统对食品所做的咀嚼功，建立了食品咀嚼性评价方法。首先以花生为试验材料，利用多因素方差分析法分析了电动机转速、仿咀嚼肌弹簧弹性系数、咀嚼循环次数对多次咀嚼破碎效果的影响，并将咀嚼模拟系统的咀嚼破碎效果和人类咀嚼效果进行对比分析。结果表明：电动机转速对咀嚼破碎效果影响不显著，仿咀嚼肌弹簧弹性系数对咀嚼破碎效果影响显著，两者交互作用对咀嚼效果影响不显著；咀嚼模拟系统的咀嚼效果和人类咀嚼效果无显著差异，可以通过调整仿咀嚼肌弹簧弹性系数使咀嚼模拟系统达到和人类相似的咀嚼效果；重复性检验差异不显著，咀嚼模拟系统咀嚼效果稳定。然后对人类咀嚼吞咽时食物团粒度进行分析，确定了花生被咀嚼至吞咽时的破碎率。结果表明：不同组花生样品在吞咽时的破碎率差异不显著，花生被咀嚼至吞咽时的颗粒大小较稳定；当花生颗粒破碎率的平均值达到91.67%时，就可认为花生颗粒团可吞咽。最后，提取达到可吞咽程度时的多次咀嚼信号曲线，计算出咀嚼信号曲线与时间轴围成的面积，将它作为咀嚼破碎食物的咀嚼做功；对花生样品进行多次咀嚼模拟测试和咀嚼性感官评价，对咀嚼功和咀嚼性感官评价结果进行相关性分析和回归分析。结果表明：咀嚼模拟系统的咀嚼功和咀嚼性感官评价结果达到了极显著的相关性；咀嚼功和感官评分建立的回归模型，满足精度要求，可以利用多次咀嚼测试的咀嚼功来测试食品的咀嚼性。综上所述，本文所研制的咀嚼模拟系统在结构上与人类咀嚼模拟系统相近，在功能上可以咀嚼破碎食物，可以用于食品质地评价；在多种咀嚼模式下建立的食品质地评价方法可以对食物的多种质地特性进行准确评价。经过与其它食品质地评价仪器对比，本文的咀嚼模拟系统不仅客观、准确，而且接近人类感官评价。更多还原

【Abstract】 Evaluation of food texture is an important aspect of food detection, has greatsignificance to food development, preservation and keeping, trade and transport, and hassome effects on human health and taking nutrition. Presently, there are two methods aboutevaluation of food texture, which are sensory assessment and instrument measurement.Sensory assessment is subjective, instabile, time consuming and laborious, while instrumentmeasurement mostly belongs to half-experience and simulation test, has much differencewith sensory assessment, and cann’t determine multiple texture characters furthermore. Withthe rapid development of food industry and increasing request of high quality food,evaluation of food texture becomes more and more important, and it needs a rapid, objective,accurate and hominine assessment method. So chewing simulation system was developed bythe technology of bionic, reverse engineering and analysis of finite element in the paper,aiming at the issue and actuality of evaluation of food texture, and multiple food texturecharacters were detected by the system under lots of evaluation patterns.1． Chewing simulation system was designed and developed including hardwareequipment and testing software. Configuration data of human chewing system was acquiredby scanning human chewing organs using Three-dimensional Laser Scanner and ComputedTomography Scanner. Chewing simulation system was developed and optimized byengineering softwares such as Imageware12, Unigraphics6.0, SolidWorks2008, Ansys10.0and so on, then was made by CAD/CAM system, including bionic teeth, bionicparodontiums, bionic jaws, bionic temporomandibular joints, driving mechanism,temperature and humidity conditions, ancillary mechanisms, signal disposing circuit etc.Testing software was designed by Visual C++6.0, including signal gathering, texture dataprocessing and figure analyzing module. After movement and chewing function of chewingsimulation system were analyzed and tested, the results show that chewing simulationsystem can realize up, down, forward, back and offset three dimensional motion alike humanchewing system, and chewing force achieve the power chewing kinds of texture foods, andmasticatory efficiency is up to90%, stable, and no significant difference with subjects’.2．Experiment conditions of the chewing simulation system were ascertained by testingnine representative texture foods. Influence that bionic parodontiums, spring imitatingchewing muscle, motor speed and temperature and humidity conditions to chewing signalwas analyzed by difference, repetition and category analysis, and the optimal experimentconditions were obtained. The suitable sensors receiving testing signal are the bionicparodontiums under and around the first bionic molar on the bionic mandible. The optimalranges of elastic coefficient to kinds of samples are different, and carrot’s is no less than30N/mm, bread’s no less than5N/mm, waxy corn, bean curd, ham sausage and crackers’ noless than10N/mm, and apple, pickle and peanut s’ no less than20N/mm. Motor speed has alittle effect on testing signal, so it should keep invariable. While carrot, peanut, apple andwaxy corn are tested, range of motor speed is60～90r/min, and while pickle, bean curd, hamsausage, bread and cracker tested, it is60～120r/min. Temperature and humidity conditions have notable influence on testing signal, so the suitable temperature is37℃, and the flow ofartificial saliva is3mL/min when the chewing simulation system chewing samples.3．Under the texture profiling analysis mode, lots of texture parameters of kinds offoods were analyzed and evaluated by chewing simulation system, sensory assessment anduniversal food texture analyzer. These foods are carrot, apple, pickle, peanut, waxy corn,bean curd, ham sausage, bread and cracker. These texture parameters are hardness,brittleness, adhesiveness, cohesiveness, springiness, chewiness and recovery property. Basedon human sensory assessment, texture profiling analysis of chewing simulation system wasset up by correlation, multivariate regression, BP network analysis, compared with universalinstrument measurement. Pearson coefficients show that the correlation is good betweenmeasurement of chewing simulation system and sensory assessment to every food samples,the chewing simulation system can replace human sense, and the correlation of chewingsimulation system testing is better than universal instrument measurement with sensoryassessment. Multivariate regression analysis shows that texture regression models on thesamples are in effect and accurate, the average error and standard residual is little and thedifference is not significant between predictive and measured value. BP network analysisshows that every BP models on the samples are effective and can satisfy with food textureevaluation.4．Food fracture mechanical model was set up based on chewing simulation. On basis ofthe “pestle and mortar” chewing model of chewing simulation system, force of materielblock was analyzed during being crushed, and fracture mechanical model was set up. Effectsof materiel block, bionic molar shape and chewing speed on the model were analyzed andcarrot, apple, pickle samples were measured to validate the model. Experiment results showthat fracture mechanical equations are correct and can express the relations between fractureforce and failure stress accurately, the fracture force of chewing simulation system canreflect the situation of materiel fracture and brittleness during being crushed. Chewingfracture force was validated and compared by sensory assessment of brittleness and threepoint bend test. Experiment results show that relations between chewing fracture force, valueof three point bend test and sensory assessment of brittleness are very marked, and thecoefficients of chewing fracture force are bigger than value of three point bend test tobrittleness sensory assessment of three samples. Regression models between chewingfracture force and sensory assessment are accurate and can predict the brittleness of crispfruits and vegetables.5．Evaluation method of food chewiness was established using the chewing work ofchewing simulation system to food, under the time after time chewing mode. Effects ofspring imitating chewing muscle and motor speed on the masticatory efficiency wereanalyzed by multiple factors variance analysis, masticatory efficiency of chewing simulationsystem was compared with conner when peanuts were chewed. Experiment results show thatmotor speed has no notable effect on masticatory efficiency, while elastic coefficient ofbionic spring has great effect, and the interaction of two factors is not effective. Masticatoryefficiency of chewing simulation system has no difference with human’s, and is stable after duplicated examination, and it can reach human’s by adjusting elastic coefficient of bionicspring. Crushing efficiency was ascertained while swallowing after analyzed granularity ofpeanut draff during human chewing. Results show that difference of crushing efficiencybetween every group is not remarkable while swallowing and granule size is stable, peanutbolus can consider to be swallowed when crushing efficiency reaches91.67%. Area thatchewing signal curves enclose time axis is as chewing work while crushing food, byextracting chewing signal curves while swallowing. After chewing simulation system testingand sensory assessment of chewiness to peanut samples, chewing work and sensorychewiness were analyzed by relation and regression. Results show that the correlation isgood between chewing work and sensory assessment and the regression model betweenchewing work and sensory assessment is accurate and can predict the food chewiness underthe time after time chewing mode.In conclusion, chewing simulation system is as human chewing system from structureto function, can chew food and evaluate food texture. Food texture evaluation method setunder lots of patterns can measure kinds of texture characters. Chewing simulation system isobjective, accurate and close to human, better than other Food Texture Analyzer aftercomparative test on food texture evaluation.更多还原