多忆阻器复合电路及其故障诊断研究

发布时间：2018-05-12 04:57

本文选题：忆阻器复合电路 + 瞬时状态　；参考：《西南大学》2017年硕士论文

【摘要】：鉴于忆阻器件的优良特性,如纳米级尺寸、非易失性、阻值可变等,有关它的研究不断增加,已有多种单个或多个忆阻器结构与先进的CMOS技术相结合,模拟实现神经形态网络中的突触或信号处理系统中的计算单元等。忆阻器在诸多领域具有应用潜能,尤其是其复合电路,但由于忆阻器的阻值变化与其极性和初始状态有关,多个忆阻器电路行为将变得复杂并难于预测。因此,为了更好理解忆阻器复合电路行为及其特性,探究不同连接方式的忆阻器电路特性是很重要的。而忆阻器的尺寸参数变化对其阻值和特性有明显影响,因而,为了提高忆阻器在各应用中的准确性和有效性,忆阻器电路的故障诊断是非常需要的,并且故障诊断是电路设计和测试中非常关键的环节。本文深入探究了多个忆阻器的复合电路特性,并分析比较了多种忆阻突触。然后,将灵敏度分析和模糊分析相结合的方法,用于诊断忆阻器复合电路和忆阻突触电路的故障。具体来说,本文的主要内容可分为如下几个部分:(1)先从物理角度,介绍了三种忆阻器模型的物理机制,包括TiO2、WOx以及Ag2S忆阻器。然后,从数学建模角度,介绍了忆阻器的两种数学建模方式,即导电通道的长度作为状态变量和导电通道的面积作为状态变量,并利用Matlab对已数学建模的忆阻器模型进行仿真实验。(2)探究了忆阻器复合电路的特性,包括串联、并联和混联结构。分析了忆阻器电路的瞬时状态和稳定状态,给出了从瞬时状态达到稳定状态的条件。然后,针对忆阻器的三种不同连接结构,进行了具体地分析,给出了不同复合结构的磁通量与电荷量之间的关系,进而得到其等效忆阻值与磁通量或电荷量之间的关系。在此基础上,研究了具有n个忆阻器的电路结构。同时,也利用相应电路结构的实验结果来说明相关推导的准确性。(3)基于忆阻器复合电路的理论推导,对比分析了多种基于忆阻器的突触。详细推导了这些突触中忆阻器阻值与输入激励的关系,进而分析了实现的突触权值是否是线性更新的,并通过仿真实验进行了验证。进一步,推导了忆阻值改变与所加外部激励的时间关系。此外,分析了厚度偏差对忆阻器特性的影响,进而探究了其对忆阻突触的影响。(4)设计了具有反馈控制的双发生器,能根据输出反馈控制开关脉冲,进而自动切断激励的输入。然后,介绍了故障诊断假说,并且基于模糊分析,推导出故障诊断假说的模糊表达式。其次,利用结点电压灵敏度分析与模糊分析相结合的方法,诊断忆阻器复合电路的故障。进一步,通过两个例子来说明此方法在忆阻器电路诊断中的有效性。
[Abstract]:In view of the excellent characteristics of the device, such as nanometer size, nonvolatile, variable resistance and so on, the research on it has been increasing, and many single or multiple memristors have been combined with advanced CMOS technology. The synapses or signal processing units in neural morphological networks are simulated. The resistor has the potential to be applied in many fields, especially its compound circuit. However, because the resistance of the resistor is related to its polarity and initial state, the behavior of the multiple circuits will be complicated and difficult to predict. Therefore, in order to better understand the behavior and characteristics of the circuit, it is very important to explore the characteristics of the circuit with different connection modes. The resistor's dimension parameter change has obvious influence on its resistance value and characteristic. Therefore, in order to improve the accuracy and effectiveness of the device in every application, it is very necessary to diagnose the trouble of the circuit. And fault diagnosis is very important in circuit design and testing. In this paper, the compound circuit characteristics of multiple amnesia are studied, and the synapses of multiple amnesia are analyzed and compared. Then, the sensitivity analysis and fuzzy analysis are combined to diagnose the fault of the compound circuit and the synaptic circuit. Specifically, the main contents of this paper can be divided into the following parts: 1) first of all, from the physical point of view, the physical mechanisms of three kinds of memristor models, including TiO2WO x and Ag2S resistor, are introduced. Then, from the point of view of mathematical modeling, this paper introduces two mathematical modeling methods of the resistor, that is, the length of the conductive channel as the state variable and the area of the conductive channel as the state variable. Matlab is used to simulate the mathematical model of the resistor. (2) the characteristics of the compound circuit are explored, including series, parallel and hybrid structure. The transient state and the stable state of the circuit are analyzed, and the conditions for achieving the stable state from the instantaneous state are given. Then, the relationship between magnetic flux and charge of different composite structures is given, and the relationship between equivalent memory resistance and magnetic flux or charge is obtained. On the basis of this, the circuit structure with n resistive devices is studied. At the same time, the experimental results of the corresponding circuit structure are used to illustrate the accuracy of the related derivation. (3) based on the theoretical derivation of the compound circuit of the memory device, the synapses based on the memory device are compared and analyzed. The relationship between the resistive resistance and the input excitation in these synapses is deduced in detail, and then whether the realized synaptic weight is linearly updated is analyzed and verified by simulation experiments. Furthermore, the temporal relationship between the change of amnesia resistance and the external excitation is deduced. In addition, the influence of thickness deviation on the characteristics of amnesia is analyzed, and its effect on the synapse of amnesia is explored. A double generator with feedback control is designed, which can control the switch pulse according to the output feedback, and then cut off the excitation input automatically. Then, the fault diagnosis hypothesis is introduced, and the fuzzy expression of fault diagnosis hypothesis is derived based on fuzzy analysis. Secondly, using the method of node voltage sensitivity analysis and fuzzy analysis, the fault of compound circuit of amnesia is diagnosed. Furthermore, two examples are given to illustrate the effectiveness of this method in the diagnosis of resistive circuits.
【学位授予单位】：西南大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TN60

【参考文献】