【作者】 席新明；

【导师】 何东健； 张乃迁；

【作者基本信息】 西北农林科技大学 ， 农业机械化工程， 2012， 博士

【摘要】 介电常数能很好地反映电介质材料的性能。介电常数测量技术的发展主要用于固体介质。大多数现有的介电常数传感器提供给定频率下的电导率或介电常数的单方面信息，只能测量一个特定范围内的单个或少数几个特征参数，不能全面了解测试样品介电参数的频率特性，因此，不能区分出介质中的多种成分。为了将介电特性传感器应用于液体和气体电介质材料的组分和特性参数检测，为水中和空气中污染物的实时检测提供一种新的方法和技术，本论文在分析液体和气体电介质特性的基础上，研究基于介电特性的流体组分检测方法与技术，用于同时测量液体和气体电介质材料的多个特性参数和多种组分，通过大量的试验研究和分析，得到不同流体组分的频率响应模式和检测模型。主要研究内容和结论包括以下几方面：（1）在分析土壤介电特性检测技术和液体、气体电介质材料特性的基础上，研究分析基于微控制器的用于同时测量流体电介质多个特性参数和识别多种组分的新型频率响应介电特性检测系统（堪萨斯州立大学生物与农业工程系测控实验室设计）的原理和检测性能。根据检测介质的特性和检测需要，选择不同材料制作传感器探头。根据试验研究需要，设计制作定位板等辅助装置。分析检测信号的主要影响因素，为试验研究中对这些因素的最佳控制提供依据。对系统的检测精度进行对比分析。（2）根据检测硬件系统的功能要求，以及试验研究中的不同需要，参考堪萨斯州立大学生物与农业工程系K.H. Lee开发的土壤检测软件，设计相应的流体介电特性检测软件系统和控制程序，选用C语言编写控制测量系统与数据采集的程序，用TaskingC166/ST10软件开发工具对程序进行编辑、调试，通过EDE和Flash Tools16W下载到kitCON167微控制器并运行。实现了产生不同频率不同数量的正弦电压信号、不同的信号等待时间、不同时间间隔下的连续自动测试取数等功能，保证了检测系统的良好运行。根据频率响应数据信息量大的特点，研究偏最小二乘（partial least square，PLS）回归、stepwise回归等数据处理方法，用Matlab设计数据处理程序，提高了数据处理的精度和效率。（3）针对生物柴油与柴油混合燃料发动机最佳动力性能发挥和排放控制中快速精确地检测燃料混合比及品质的要求，提出了一种根据介电特性快速、准确地测量柴油/生物柴油混合比的新方法。采用流体介电特性检测系统，设计试验研究生物柴油和柴油的混合比及燃油品质与其介电特性的关系。分析生物柴油混合燃料的频率响应模式，用PLS和stepwise回归法进行拟合，建立混合比的检测模型。结果表明，频率响应特性与混合比之间具有很高的相关性，PLS回归的训练与验证模型的相关系数R~2都大于0.99，而相应的均方根误差RMSE（root-mean-square error）都小于0.9%。stepwise回归使数据量显著减少，也有较高的预测精度。采用介电常数和频率响应法测量生物柴油的混合比可以得到高精度，而且能满足快速测量的要求。（4）针对混合燃料车辆发动机最佳点火时间调整和排放控制对精确、快速、方便地测量燃料混合比和质量参数的要求，采用流体介电特性检测系统，设计试验研究乙醇/汽油混合燃料的混合比、含水量与其介电特性之间的关系，分析乙醇/汽油混合燃料的频率响应模式。采用偏最小二乘（PLS）和stepwise回归，构造混合比的检测模型。PLS回归的训练与验证模型的相关系数R~2都大于0.99，而相应的均方根误差RMSE都小于1.2%。研究结果表明，使用介电性能和频率响应法检测生物汽油的混合比可以得到高精度，并能满足快速测量的要求，为生物汽油混合比的实时、准确测量奠定了基础。（5）针对环境安全和人类健康保证对高效率、低成本水质实时监测技术的迫切需求，采用流体介电特性检测系统，设计大量试验，研究钾盐、硝酸盐水溶液中不同离子浓度与其介电特性的关系，分析不同离子的频率响应模式，研究基于介电特性的水中钾盐、硝酸盐、除草剂等污染物浓度的快速、准确的检测方法。采用PLS回归，建立钾盐和硝酸盐水溶液中各种离子浓度的预测模型。通过主成分分析，识别钾离子和硝酸根离子的特征频率，用特征频率建立预测水中钾离子浓度和硝酸盐离子浓度的新模型。结果表明，每种离子都有其独特的频率响应模式，不同种类的盐溶液其频率响应特性在低频区存在明显差异。PLS回归模型对低浓度混合溶液中的钾离子浓度和硝酸根离子浓度都显示了很高的预测精度。主成分分析和特征频率的识别使数据量大幅度减少，同时也有较高的预测精度。本研究为高效率、低成本的水中污染物实时监测技术奠定了基础。（6）针对传统空气污染物测量技术中存在的采样时间长、前处理步骤繁琐、待分析的挥发性物质易损失，以及测量费用高和连续测量能力缺乏等问题，采用流体介电特性检测系统，对空气中颗粒污染物、柴油车尾气和挥发性有机化合物等几种类型的空气污染物检测进行初步研究和探索。结果表明，空气样本的频率响应模式的变化，主要出现在中高频区。电介质的电容作用在空气中占主导地位，而空气中的导电是非常小的。（7）分析了测试系统的优缺点，提出了今后进一步深入研究的思路。更多还原

【Abstract】 Permittivity can well reflect the performance of dielectric materials. Permittivitymeasuring techniques have been developed mainly for solid materials. Most existingpermittivity sensors provide only one piece of information–apparent EC or dielectricconstant at a given frequency, only measuring single or a few characteristic parameters withina particular range, and are unable to fully uederstand the frequency characteristics of thedielectric parameters of the test sample. Thus, these instruments are incapable ofdistinguishing multiple ingredients that influence the permittivity in dielectric materials.In order to apply permittivity sensors to measurement of composition and properties ofliquid and gas dielectric materials, and to provide a new method and technique for real-timedetection of water and air pollutants, in this paper, the detecting method and technology offluid compositions based on permittivity have been studied to detect multiple properties offluid dielectric simultaneously and identify a variety of compositions on the basis ofanalyzing the properties of liquids and gases dielectric. And through a large number ofexperimental research and analysis, the frequency-response pattern and detecting models ofdifferent compositions in fluid have been obtained.The main contents and conclusions in this study are including as follows:1. On the basis of analysis detection technology of soil dielectric properties and materialproperties of liquid and gas dielectric, the principle and the detection performance of the newfrequency response permittivity detection system based on microcontroller developed by theInstrumentation and Control Laboratory of Biological and Agricultural EngineeringDepartment of Kansas State University used to detect multiple properties of fluid dielectricsimultaneously and identify a variety of compositions have been researched and analized. Thesensor probes have been made selecting different materials according to the characteristics ofthe detection medium and testing need. According to the needs in test and research, thepositioning plate has been designed and made.The main factors of the detection signal havebeen analyzed so as to achieve optimum control on these factors in test study. The detectionaccuracy of the system was analyzed by contast.2. According to functional requirements of detection hardware system, and differentneed in test and research, the corresponding fluid permittivity detection software system and controngling programs have been designed reference to the Soil Detecting Softwaredeveloped by K.H. Lee of Biological and Agricultural Engineering Department of KansasState University. Programs for measurement system controlling and data acquisition werewritten and compiled using C language. The programs were edited and debugged withsoftware developing tool Tasking C166/ST10， and were downloaded to kitCON167microcontroller for running through EDE and Flash Tools16W. Functions to producesinusoidal voltage signals with different number and different frequencies, different waitingtime of signal, colletting data continuously and automatically under different time intervals etal. have been achieved. The detection system has been ensured for a good run. Based on thefeature of frequency response data with large amount of informations, the data processingmethods of partial least squares (PLS) and stepwise regression have been studied. Dataprocessing program has been designed using Matlab software, and the accuracy and efficiencyof data processing have been improved.3．In connection with the requirements of detection the fuel mixing ratio and quality fastand accurately for exerting the best power performance and exhaust emission control to blendfuel engine of biodiesel and diesel, a new method to detect the mixing ratio of diesel/biodiesel blends rapid and accurately based on the dielectric properties was proposed. Thepermittivity detection system developed for fluid dielectric materials was used to study on therelationship of blend ratio and frequency-response characteristics of biodiesel and dieselblends through experiment. The frequency-response pattern of biodiesel blend fuel has beenanalyzed. With PLS and stepwise regressions, the blend ratio detecting models have beenconstructed. The results show that correlation of mixing ratio and frequency-response are veryhigh. R-square values are greater than0.99while RMSE are less than0.9%for both trainingand validation of PLS regression models. The use of stepwise regression method significantlyreduces the amount of data needed to measure the bend ratio, also have a high accuracy.Using the permittivity and frequency response method to measure the mixing ratio ofbiodiesel blend can get high precision and can meet the requirement of fast detecting.4. In connection with the requirements of detection the fuel mixing ratio and quality fast,accurately and conveniently for adjusting the engine’s optimal ignition time and exhaustemission control to engine of Hybrid vehicle, the permittivity detection system developed forfluid dielectric materials was used to study on the relationship between mixing ratio, watercontent and dielectric characteristics of ethanol/gasoline blends through experiment. Thefrequency-response pattern of ethanol/gasoline blend fuel has been analyzed. With PLS andstepwise regressions, the mixing ratio detecting models have been constructed. R-square values are greater than0.99while RMSE are less than1.2%for both training and validationof PLS regression models.Study results show that using the dielectric property and frequencyresponse method to measure the mixing ratio of biogasoline can get high precision and canmeet the requirement of fast detecting, which laid the foundation for real-time and accuratemeasurement of biogasoline mixing ratio.5. In connection with the urgent needs for real-time monitoring technology to waterquality with high efficiency and low cost so as to protect environmental safety and humanhealthy, the permittivity detection system developed for fluid dielectric materials was used tostudy on the relationship between concentration of different ion or cation in potassium，nitratesalt solutions and dielectric properties by large number of experiments. Thefrequency-response pattern of different ion or cation has been analyzed. Study was done onthe rapid and accurate method of concentration detection of potassium salt, nitrates, herbicideresidual and other pollutants in water based on the dielectric properties. With PLS regression,the concentration prediction models of various ion or cation in potassium salt and nitratesolutions have been established. Based on a principal component analysis, characteristicfrequencies for potassium cation and nitrate ion were identified from the frequency-responsepatterns. The new prediction models at the characteristic frequencies for concentration ofpotassium cation and nitrate ion in water have been established. The results showed that eachion or cation has its unique frequency-response pattern; frequency-response properties ofdifferent types of salt solution are significant difference at the low frequency region.Prediction accuracy of PLS regression models for potassium cation concentration and nitrateion concentration in mixed solution with low concentration are all very high. Principalcomponent analysis and identification of the characteristic frequencies reduce the amount ofdata significantly, and also have a high accuracy. The study laid the foundation for real-timemonitoring technology of water pollutants with high efficiency and low cost.6. In connection with the issues of long sampling time, tedious pre-processing step,volatile substances for analysis to be easily lost, and the high cost of measurement and lack ofcontinuous measurement capacity which present in traditional air pollutants measurementtechniques, the permittivity detection system developed for fluid dielectric materials was usedto make a preliminary study and probing on detection of several types of air pollutants such asparticulate air pollutants, diesel exhaust, and volatile organic compounds. Preliminaryexperimental results show that air samples changes in the patterns of frequency response,mainly in the middle and high frequency region. In most of the air samples, the role ofdielectric capacitance dominates, and the conductivity of the air is very small. 7. Advantages and disadvantages of the frequency response sensor were analyzed.Further studies to improve the sensor performance were recommended.更多还原