环境对烟叶品质的影响分析及霉变预测方法研究

发布时间：2018-07-13 18:23

【摘要】：烟草(Nicotiana tabacum L.)属管状花目,茄科一年生或有限多年生草本植物。烟草的分布范围遍及北纬60度到南纬45度间的绝大部分国家和地区。中国是世界烟叶生产第一大国,常年种植烟100多万公顷,烟叶年产量达到200多万吨。在我国,烟草种植分布于绝大部分省份和地区,同时烟草也是重要的经济作物,烟草行业的税收是我国财政收入的重要组成部分。由于烟草对生态环境的适应性较强,所以在世界多地广泛种植,但作为叶用植物,其烟叶的长相、组织结构、物质转化与积累状况等因素直接影响质量。从烟叶质量角度看,烟草又对生态环境的变化十分敏感。生态环境的差异及栽培措施的不同,不仅影响烟株形态和农艺性状,还导致烟叶内在化学成分的变化,进而影响烟叶品质。长期以来,人们从生态环境、品种、栽培措施、调制等与烟叶质量的关系方面开展了大量研究,并提出了生态环境、品种、栽培技术对烟叶质量的贡献率分别为56%、32%、10%左右的说法。生态环境是决定烟叶品质最重要的因素,主要体现在生态环境影响烟叶香气物质含量和组成比例,进而主导烤烟香气风格,并使烟叶香气和吃味彰显典型的地域特征等方面。生态因素中,气候因子对烟叶香型和感官质量指标的影响及贡献率最大,人们围绕气候因子与烟叶香气物质的关系开展了系列研究,但大多局限在某一特定地区的单一气候因子与烤烟香气物质之间关系研究。常见的气候因子主要包括光照、湿度、降水量、温度等。本论文以烟叶品质(包括外观质量、物理特性、感官质量、化学成分)为研究对象,通过系统分析烟叶生产过程中的大田生长环境(海拔、气温、湿度、降雨量和光照)和仓储环境(温度、湿度、微生物)两个阶段中环境因子对烟叶品质的影响,建立非线性模型,并将此方法应用于不同环境条件下烟叶的品质预测研究。(1)烟叶品质特性的分析研究。从化学成分和感官质量两个不同方面对不同环境条件下烟叶的风格类型和品质特征进行表征,并以感官评价风格特征指数为导向,挑选代表性化学指标体系：与品质特征指数相关的代表性化学指标为总糖、总氮、钾、多酚总量、茄酮、巨豆三烯酮A、香叶基丙酮、苯乙醇、辛酸；与风格特征指数相关的代表性化学指标为总糖、总氮、钾、多酚总量、巨豆三烯酮D、藏花醛、香叶基丙酮、苯乙醛、丙二酸、苹果酸。(2)种植环境对烟叶品质的影响研究。采用相关性分析、多重比较分析、偏最小二乘判别分析、回归分析及神经网络分析等方法指标筛选受环境因子影响的烟叶品质指标。受海拔因子影响较的指标主要有清甜香、坚果香、清香型、中间香型、浓香型、风格指数、柠檬酸、棕榈酸、亚油酸、异戊酸、2-甲基-丁酸、挥发性酸总量、叶黄素、p-胡萝卜素、芸香苷、醇类和酸类共计17项指标。通过回归分析可知,上述17项指标受低、中、高海拔的影响,交叉验证正确率分别为：77.8%,83.0%和85.0%。受年平均气温影响的指标主要有：浓度、清甜香、坚果香、焦甜香、木香、蜜甜香、清香型、烟碱、绿原酸和醇类共计12项指标,通过回归分析可知,上述12项指标受低、中、高气温的影响,交叉验证正确率分别为：80.00%,81.25%和80.00%。受年平均湿度影响的指标主要有：还原糖、总氮、钾、柠檬酸、棕榈酸、2-甲基-丁酸、叶黄素、β-胡萝卜素、绿原酸、多酚总量和醇类共计12项指标,通过回归分析可知,上述12项指标受低、中、高湿度的影响,交叉验证正确率分别为：83.3%,92.9%和91.7%。受年平均降雨影响的指标主要有：木香、烟碱、总糖、钾、苹果酸、亚油酸、非挥发性酸总量、苯甲酸、辛酸、叶黄素、β-胡萝卜素、莨菪亭、芸香苷、多酚总量、醛类和杂环类共计16项指标,通过回归分析可知,上述16项指标受低、中、高降雨的影响,交叉验证正确率分别为：83.8%,86.6%和83.3%。受年平均日照影响的指标主要有：香气量、浓度、杂气、干净度、湿润、品质指数、清甜香、干草香、坚果香、蜜甜香、中间香型、风格指数、戊酸、己酸、苯甲酸、绿原酸和芸香苷共计17项指标,通过回归分析可知,上述17项指标受低、中、高日照的影响,交叉验证正确率分别为：100.0%,83.3%和83.3%。为进一步研究环境因子(海拔、年平均气温、年平均降雨、年平均湿度和年平均日照)共同对烟叶品质指标的影响,采用神经网络分析优化受环境因子影响较的指标,通过BP网络通过多次迭代分析,建立了清香型、浓香型和清甜香指标的预测模型,该3项指标训练集相关系数分别为：98.16%、96.89%、85.52%和79.29%,且测试集效果较好,可用于上述三个指标的预测。通过本章系统研究环境因子对烟叶品质指标的影响,这为后续进一步研究环境因子对烟叶生理生化、生长发育的影响奠定了基础。(3)烟叶仓储环境中的霉菌及其对烟叶品质的影响研究。首先进行霉菌微生物的分离、纯化及致腐性测定,筛选出烟叶中的导致霉变的主要霉菌,并通过微生物的培养及形态学鉴定判断霉菌种属,本研究烟叶中的霉菌主要为曲霉属和青霉属。对霉变烟叶的化学成分进行分析。通过对烟叶营养成分的分析,表明霉变烟叶样品中C源、N源和矿物质类化学成分均有不同程度的变化。糖类、淀粉、纤维素、木质素、有机酸、多酚等C源类物质含量降低幅度较大,氯离子、硫酸根、硝酸根和磷酸根等阴离子元素含量基本没有变化；通过霉变烟叶挥发性及半挥发性成分分析,表明霉变烟叶中新产生的化合物有丁醇、2-甲基-丁醇、戊醇、己醇、亚油酸乙酯、亚麻酸乙酯、硬脂酸乙酯。含量变化程度较大的化合物：棕榈酸乙酯、棕榈酸甲酯、3-甲基-丁醇、4-甲基苯酚、3-羟基-2-丁酮、5-甲基糠醛、吲哚、亚麻酸甲酯、苯乙醇、糠醛、苯甲醇和糠醇。应用筛选出的这19种烟叶霉变的挥发性特征成分建立模型,正常样品判别模型的预测准确率达95.0%、霉变样品的预测准确率达94.4%。预测准确率可满足实际应用,应用本方法可实现烟叶霉变的有效判定。在系统分析烟叶霉变的化学成分变化及霉变程度的判定方法基础上,应用近红外光谱技术进行烟叶霉变预测模型的建立与验证和麦角甾醇定量预测模型的建立与验证。应用近红外光谱仪在780 nm-2500 nm范围内对烟叶的近红外光谱进行采集,获得烟叶样本的基础光谱数据。建立GBA算法,对基础光谱数据进行特征波长的筛选,并建立PLS-DA判别模型。判别模型具有较小的Wilks'λ0.216 (P0.001)和误差率2.92%。将该模型应用于烟叶样品的识别,准确率高达95.79%。尤其是,可以在烟叶霉变的I期得到满意的识别,适用于烟叶霉变的预测及仓储烟叶的实时监控。在烟叶出现霉变或者有出现霉变趋势的情况下,及时采取通风、降温、除湿、翻垛、打叶复烤降低含水率等措施,防止霉变的进一步扩散和蔓延,最大可能的降低损失。应用近红外光谱法对烟叶中的麦角甾醇含量进行预测,提供一种检测速度快、操作简单方便、预测结果准确的快速测定的方法,对于烟叶霉变的预测具有实际意义。
[Abstract]:Nicotiana tabacum L., an annual or Limited perennial herb of the family Solanaceae. The distribution range of tobacco is in the vast majority of countries and regions between 60 degrees north latitude and 45 degrees south latitude. China is the world's largest tobacco production country, growing about 1000000 hectares of tobacco in a year and the annual yield of tobacco to about 2000000 tons. In China, tobacco Planting is distributed in most provinces and regions, and tobacco is also an important economic crop, and the tax revenue of the tobacco industry is an important part of our financial income. Because tobacco has a strong adaptability to the ecological environment, it is widely cultivated in the world, but as a Ye Yongzhi, the length, structure, material transformation and product of the tobacco leaves. From the point of view of the quality of tobacco, tobacco is very sensitive to the changes in the ecological environment. The differences in the ecological environment and the different cultivation measures not only affect the tobacco plant morphology and agronomic traits, but also change the chemical composition of the tobacco leaves, and then affect the quality of tobacco. For a long time, people have from the ecological environment, A lot of studies have been carried out on the relationship between varieties, cultivation measures and modulation and the quality of tobacco leaves. The contribution rate of ecological environment, variety and cultivation technology to tobacco quality is 56%, 32%, and 10% respectively. The ecological environment is the most important factor in determining the quality of tobacco leaves, which is mainly reflected in the effects of the ecological environment on the aroma content of tobacco leaves. In the ecological factors, the influence and contribution rate of climate factors on tobacco flavor and sensory quality indexes are the greatest. The relationship between a single climate factor in a particular area and the aroma substances of flue-cured tobacco. The common climatic factors mainly include light, humidity, precipitation, and temperature. In this paper, the quality of tobacco (including appearance quality, physical properties, sensory quality, chemical composition) was studied by systematic analysis of field growth in the process of tobacco production. The environment (altitude, temperature, humidity, rainfall and illumination) and storage environment (temperature, humidity, microorganism) in two stages of environmental factors on the quality of tobacco leaf quality, establish a nonlinear model, and apply this method to the quality prediction of tobacco under different environmental conditions. (1) analysis of the quality characteristics of tobacco leaves, from chemical composition and sensory quality The style and quality characteristics of tobacco leaves under different environmental conditions were characterized by two different parties, and the representative chemical index system was selected based on the sensory evaluation style index. The representative chemical indexes related to the quality characteristic index were total sugar, total nitrogen, potassium, total polyphenols, Solanone, soybean three ketone A, and fragrant leaf base C Ketone, benzol, and octanic acid; the representative chemical indexes related to the style characteristic index are total sugar, total nitrogen, potassium, polyphenols, three ketone D, Tibetan aldehyde, fragrant leaf acetone, phenylacetaldehyde, malonic acid, malic acid. (2) study on the effect of planting environment on tobacco quality. Correlation analysis, multiple comparison analysis, partial least squares discriminant analysis, and recovery The indexes of tobacco leaf quality influenced by environmental factors were selected by regression analysis and neural network analysis. The main indexes affected by altitude factors were sweet fragrance, nut aroma, fragrance type, middle flavor type, Luzhou flavor, style index, citric acid, palmitic acid, linoleic acid, isovaleric acid, 2- methyl butyric acid, total volatile acid, lutein, p- carrot. According to the regression analysis, the above 17 indexes were affected by low, middle and high altitude, and the accuracy of cross validation was 77.8%, 83% and 85.0%. were mainly influenced by annual mean temperature: concentration, sweet fragrance, hard incense, coke sweet, wood aroma, honey sweet, fragrance, nicotine, chlorogenic acid. A total of 12 indexes and alcohols, through regression analysis, the above 12 indexes were affected by low, middle and high temperature. The accuracy of cross validation was 80%, 81.25% and 80.00%. were mainly affected by annual average humidity: reducing sugar, total nitrogen, potassium, citric acid, brown acid, 2- methyl butyric acid, lutein, beta carotene, chlorogenic acid, polyphenols. Total and alcohols were 12 indicators. Through regression analysis, the above 12 indexes were affected by low, medium and high humidity. The accuracy of cross validation was 83.3%, 92.9% and 91.7%. were mainly affected by annual rainfall: wood, nicotine, total sugar, potassium, malic acid, linoleic acid, total non volatile acid, benzoic acid, octanoic acid, lutein, beta - carotene, scopolamine, rutin, rutin, polyphenols total, aldehydes and heterocyclic 16 indexes. Through regression analysis, the above 16 indexes are affected by low, middle and high rainfall. The accuracy of cross validation is 83.8%, 86.6% and 83.3%. are mainly affected by annual average sunshine: aroma, concentration, impurity, cleanliness, wetness, and quality. Quality index, sweet scent, hay fragrance, nut fragrance, honey sweet, middle flavor type, style index, valerate, hexanic acid, benzoic acid, chlorogenic acid and rutin are 17 indexes. Through regression analysis, the above 17 indexes are affected by low, medium and high sunshine, and the accuracy of cross validation are 100%, 83.3% and 83.3%. for further study of environmental factors (SEA). The effects of annual average temperature, annual average rainfall, annual average humidity and annual average sunshine on the quality index of tobacco leaf were combined, and neural network was used to optimize the index of the influence of environmental factors. Through multiple iterative analysis of BP network, the prediction model of scent, Luzhou flavor and sweet scent was established, and the 3 indexes were trained. The number of relations is 98.16%, 96.89%, 85.52% and 79.29% respectively, and the test set effect is good. It can be used for the prediction of the above three indexes. Through this chapter, the influence of environmental factors on the quality of tobacco leaves is studied, which lays a foundation for further research on the effects of environmental factors on the physiological and biochemical, growth and development of tobacco. (3) the storage environment of tobacco leaves Study on the mold and its effect on the quality of tobacco. First, the isolation, purification and rot induced determination of fungi were carried out, and the main moulds in the leaves were screened out, and the fungi were judged by the culture of microbes and the morphological identification. The mold in the tobacco leaves was mainly Aspergillus and Penicillium. The chemical composition of the tobacco leaf was analyzed. Through the analysis of the nutrient composition of the tobacco, it was shown that the C source, the N source and the chemical composition of the minerals in the moldy tobacco leaves varied in varying degrees. The contents of C sources such as carbohydrates, starch, cellulose, lignin, organic acids and polyphenols were greatly reduced, and chloride, sulfate, nitrate and phosphate were removed. The content of the subelements is basically unchanged. Through the analysis of the volatile and semi volatile components of mouldy tobacco, it is shown that the newly produced compounds in the mouldy tobacco are butanol, 2- methyl butanol, pentanol, hexanol, ethyl linolenic acid ethyl ester, and ethyl stearate. The content of the compounds with a large change in content: ethyl palmitate, palmitate methyl ester, 3- a Base butanol, 4- methyl phenol, 3- hydroxy -2- butanone, 5- methyl furfural, indole, linolenic acid methyl ester, benzyl alcohol, furfural, benzyl alcohol and furfuryl alcohol. The model was established to determine the volatile characteristics of the 19 kinds of tobacco leaves. The prediction accuracy of the normal sample model was 95%, and the prediction accuracy of the mouldy samples reached the accuracy of 94.4%. prediction. It can be applied to the practical application. This method can be used to realize the effective determination of the leaf mould. On the basis of the systematic analysis of the chemical composition and the degree of mildew, the near infrared spectroscopy is used to establish and verify the mould prediction model and the establishment and verification of the model of the quantitative prediction of ergosterol. The infrared spectrum of the tobacco leaves is collected in the range of 780 nm-2500 nm to obtain the basic spectral data of the tobacco leaf samples. The GBA algorithm is established to screen the characteristic wavelengths of the basic spectral data and establish the PLS-DA discriminant model. The discriminant model has a smaller Wilks'lambda 0.216 (P0.001) and an error rate 2.92%. to apply the model to the model. The accuracy of the tobacco leaf sample is up to 95.79%., especially, it can be recognized satisfactorily in the I period of the leaf mold. It is suitable for the prediction of the leaf mould and the real-time monitoring of the storage leaf. Measures to prevent further spread and spread of mildew, the maximum possible reduction of loss. Using near infrared spectroscopy to predict the content of ergosterol in tobacco leaves, providing a rapid detection method, simple and convenient operation, accurate prediction results, for the prediction of tobacco moldy is of practical significance.
【学位授予单位】：昆明理工大学
【学位级别】：博士
【学位授予年份】：2016
【分类号】：S572

【参考文献】