【作者】 吴云飞；

【导师】 叶齐政；

【作者基本信息】 华中科技大学 ， 电气工程， 2013， 博士

【摘要】 介质阻挡放电(DBD)是一种典型的能够在常温常压下产生大量高能量密度的非平衡等离子体源，目前已被广泛应用于臭氧制备、高功率激光器、等离子体流动控制等诸多领域。由于工业应用的需要，放电的均匀性一直是研究的热点问题。目前所谓均匀放电仅指的是短时间尺度上（ns级）无任何微放电细丝的绝对均匀放电，但实际上工业应用并不需要限制在这种绝对均匀的放电上。相反，更为需要的是由大量微放电细丝构成的长时间尺度上相对均匀的放电。由此引出这种放电的均匀性如何定量评价的问题，如何产生不同均匀性的放电及提高均匀性的问题，以及均匀放电和绝对均匀放电的物理机制问题。本文以DBD均匀性为主要研究对象，通过采用数字图像处理技术并结合电气参数测量手段及近似解析计算方法对其开展了研究，所取得的成果如下：提出了一种利用数字图像处理技术中的灰度直方图（GLH）来定量识别DBD均匀性的新方法。该方法克服了传统方法复杂、繁琐及价格昂贵等缺点，能够简单和有效地识别丝状和均匀放电模式并定量评价放电均匀性。GLH方法的有效性得到了能产生绝对均匀放电的低气压实验和一般采用的电流波形区别方法的验证。在此基础上，通过采用信赖域算法求解非线性最小二乘问题得到双高斯及单高斯概率模型的参数，建立了丝状和均匀放电的灰度概率模型并将该模型应用于DBD放电特性的研究。提出了通过数字图像处理技术中的傅里叶能量谱（FES）、空间自相关函数（ACF）及灰度共生矩阵（GLCM）来定量识别DBD图像空间结构的方法。ACF方法具有噪声抑制能力强、对象识别能力强、不需要任何预处理过程等明显优点，有效识别了丝状、周期性及均匀放电图像的空间结构。该方法的有效性通过对DBD斑图空间结构的识别得到验证。研究了丝网电极对DBD均匀性的影响。（1）实现了具有空间周期特征丝网电极的均匀放电，部分特征尺度下的均匀性要好于平板电极。实验结果表明：如果丝网电极的孔径L足够大（L≥1.5mm），周期性的放电点将在每一个网格点上出现；如果L相对较小（1.25mm≥L≥0.6mm），周期性的放电点将交替出现；如果L≤0.5mm，放电点将随机分布在电极表面，且其密集程度会超过平板电极，其中放电点没有产生在每一个网格点上的原因与电子崩尺寸大小有关。FES及ACF分析验证了上述实验结果。（2）提出了定量评价DBD均匀性的指标——图像的灰度变异系数（CV）：CV越小，放电越均匀，反之，则越不均匀。研究结果表明，在一定程度上，随着L的减小，CV逐渐减小，放电均匀性逐渐增加。（3）提出了控制丝网放电均匀性的标度不变量h来分析导致相对均匀DBD的物理机制,其中h包含了L及归一化的电子崩头部电场变化率的综合效应。研究结果表明，不同丝网电极h的变化趋势和CV的变化趋势基本一致，意味着采用h可以从理论上较好地解释导致丝网电极放电均匀性的物理机制。研究了气体成分对DBD均匀性的影响。（1）发现采用GLH方法可以识别不同性质气体放电的均匀性。研究结果表明，非惰性气体（空气和氮气）放电图像对应的GLH及CV与外电压的关系曲线与惰性气体（氦气和氩气）存在明显不同，前者分别具有更大的半高宽及产生了明显的向右偏移。（2）发现氮气中加入氩气有利于提高放电均匀性。CV计算结果表明，空气、氮气、氦气、氮气/氩气混合气体及氩气放电图像对应的CV依次减小，放电均匀性依次增加。（3）DBD微放电扩展等效电气模型的计算结果表明，具有较小击穿电压的气体，所产生的微放电更容易扩展到整个电极表面，越容易形成均匀放电。另外，由于氩原子亚稳态能量明显高于氮分子亚稳态的能量，氮气中氩气的加入使得混合气体更容易通过彭宁电离产生种子电子，因此相比氮气更容易形成均匀放电。研究了均匀DBD形成的物理机制。（1）在综合考虑电子崩头部扩散和静电排斥作用的基础上，建立了多电子崩径向扩展动力学模型，提出了一个与外电场、气压、电子温度、预电离非均匀性及阀值有关的形成均匀放电所需最小种子电子密度的表达式,相关实例验证了该模型的有效性。（2）在此基础上，对单一气隙介质及DBD中的电子崩发展过程进行了近似解析计算。结果表明，当电子崩发展到一定阶段时，崩头的扩展机制将由扩散扩展过渡到静电排斥扩展。另外，通过提高种子电子数以减小电子崩的初始间距将有利于实现相邻电子崩之间的耦合，从而提高了放电均匀性。上述结果较好地验证了电子崩径向扩展动力学模型的正确性，同时也有助于加深对均匀DBD物理机制的理解。更多还原

【Abstract】 Dielectric barrier discharge (DBD) is a typical non-equilium plasma source atatmospheric pressure, which has been extensively used for various industrial applications,such as ozone production, high power lasers and the plasma flow control. The uniformity ofdischarge has been a research focus due to the needs of industrial applications. Now theso-called uniform discharge only refers to the absolute uniform discharge without anymicrodischarges in a short timescale (ns). But in fact the uniform discharge used inindustrial applications does not need to be limited to the absolute uniform discharge. On thecontrary, the relative uniform discharge consisted by a large number of microdischarges ina long timescale may meet the needs of more industrial applications. Therefore, theproblems on how to evaluate this discharge uniformity quantitatively, how to produce thedischarge with different uniformity, how to improve discharge uniformity, and the physicalmechanisms of uniform and absolutely uniform discharge are raised. This paper studied theuniformity of DBD by using digital image processing technology, electrical measurementsand approximate analytical calculation method, the research results are as follows:A new method—gray level histogram (GLH) based on the digital image processingtechnology was proposed to classify the uniformity of DBD quantitatively. Thedisadvantages of complex, cumbersome and expensive can well be overcome by using theGLH method instead of the conventional method. So it can be used to classify thefilamentary and uniform discharge and evaluate the discharge uniformity quantitatively in asimple and effective way. In addition, the effectiveness of the GLH method in classifyingthe different discharge modes was validated by the experiments under low pressure whichcan produce absolutely uniform discharge and the distinguishing method of currentwaveform which commonly used by researchers. Moreover, the gray level probabilitymodel of filamentary and uniform discharge were established by obtaining the parametersof double Gaussian and single Gaussian probability model, which were obtained by solvingnonlinear least squares problems using trust region algorithm. Furthermore, the model wasapplied to study the discharge characteristics of DBD.Fourier energy spectrum (FES), autocorrelation function (ACF) and gray levelco-occurrence matrix (GLCM) method based on the digital image processing technologywere proposed to identified quantitatively the spatial structure of the discharge image in DBD. The spatial structure of filamentary, periodic and homogeneous discharge wereidentified effectively by using thev ACF method for the obvious advantages of suppressingthe noise, high object recognition ability and requiring not any preprocessing. Theeffectiveness of the ACF method was validated by identifying the spatial structure ofpatterns in DBD.The influences of mesh electrodes on the uniformity of DBD were studied.(1) Theuniform discharge was produced by using the mesh electrodes with a characteristic ofspatial period, and the discharge uniformity produced by some mesh electrodes with certaincharacteristic lengths was better than plate electrode. If the aperture of the mesh electrodewas long enough (L≥1.5mm), those periodic discharge spots will be produced on every gridnode of the mesh electrode. If the aperture was slightly smaller (1.25mm≥L≥0.6mm), thoseperiodic discharge spots will not be produced on every grid node, but on alternate gridnodes. If L≤0.5mm, the discharge spots will be distributed randomly and even more denselythan when produced by a planar electrode.The reason that these periodic spots were notproduced on every grid node should be related to the size of an avalanche. Theexperimental results mentioned above were validated by analysis of the FES and ACF.(2)A coefficient of variation (CV) of discharge image was proposed to evaluate the uniformityof the discharge qualitatively. The smaller the CV, the more uniform the discharge will be.The research results show that the CV decreases gradually as the aperture of the electrodedecreases to a certain extent, thus the uniformity increases.(3) A new dimensionless scaleinvariant (h) was introduced to analyze the physical mechanisms leading to the relativelyuniform DBD, which includes the combined effect of the aperture and the normalizedchange rate of the field strength of the avalanche head. The research results show that thevariations of CV and h for the different electrodes have almost the same trends. This meansthat the physical mechanisms leading to the uniform discharge in mesh electrode can bewell explained by using the h.The influences of gaseous species on the uniformity DBD were studied.(1) It wasfound that the discharge uniformity in different types of gases can be classified by using theGLH method. The research results show that the GLH and the dependence of the CV onapplied voltage for non-noble gases (Air, N2) were much different from the noble gas (He,Ar).(2) The addition of Ar to N2can help to improve the discharge uniformity. Thecalculated results of the CV for different gases show that the CV decreases in the order Air,N2, He, N2/Ar gas mixture and Ar, thus the discharge uniformity increases successively.(3)The calculation results of the equivalent electrical model of microdischarge expand in DBD show that the gas with smaller breakdown voltage can make the microdischarge easier toexpand the entire electrode surface and form uniform discharge. Moreover, the energy ofatomic metastables in argon was significantly higher than the energy of moleculesmetastable in nitrogen, so more seed electrons can be produced through penning ionizationwhen the addition of Ar to N2, which was beneficial to forming uniform discharge in N2/Argas mixture.The physical mechanisms leading to the uniform DBD were studied.(1) A multipleelectron avalanches radial expansion dynamic model was established by considering theeffects of electron diffusion and electrostatic repulsion in the avalanche head. Theminimum required preionization level for the formation of multiple electron avalanchescoupling was found to be dependent on electric field strength, gas pressure, electrontemperature, the heterogeneity of preionization level and threshold value. Moreover, theeffectiveness of the model was validated by a relevant application example.(2) Theapproximate analytical calculation on the development of electron avalanches in pure gasdielectric and DBD were carried out. The results show that the expansion mechanisms inthe avalanche head will transit from free electron diffusion to electrostatic repulsion whenthe electron avalanche develops to a certain stage. Moreover, reducing the initial spacingbetween electron avalanches by improving the seed electron is beneficial to coupling theadjacent electron avalanches, thus improving the discharge uniformity. The above resultsvalidate the self-consistency of the electron avalanche radial expansion dynamic model,which can help us better understand the physical mechanisms leading to the uniform DBD.更多还原