【作者】 王嘉家；

【导师】 彭敏放；

【作者基本信息】 湖南大学 ， 电气工程， 2010， 硕士

【摘要】 模拟电路故障诊断技术一直是电路与系统领域的研究热点,然而由于电路的规模化、系统化和封装设计,造成诊断适用的测试数据过少,即使采用多类测试信号结合高效的信号处理方法,也很难从根本上解决诊断信息不足的问题,从而制约了模拟电路故障诊断技术的进一步发展。考虑到在系统中传播的故障信息一方面是故障诊断的重要信息源,另一方面,也会对真正故障源的识别与定位产生严重干扰,造成故障识别与定位的艰巨性,因此对故障信号的跟踪以及故障传播特性这一深层诊断知识进行分析就显得尤为重要。为此,本论文致力于研究大规模模拟电路故障传播特性的分析方法、仿真建模及其在故障诊断中的应用。考虑到大规模电路的故障传播特性与其拓扑结构的密切关系,本文首先以测试基准电路为例验证了大规模电路的小世界和无尺度特征,并进一步探索了大规模电路系统的层次划分方法及其实现。同时,针对模拟电路故障划分的不确定性,采用云理论来进行电路故障模式的划分以及故障初始样本集的扩展,在此基础上,提出了基于数据场的模拟电路故障传播行为模拟方法。该方法可以较好地描述故障模式之间的耦合关系,并刻画电路故障的传播情况。为建立模拟电路故障传播特性分析的有效模型,本文将图论、模糊论和概率论与复杂网络相结合,提出了故障传播特性协同仿真和建模方法,从所建模型的各种参数可获悉模拟电路故障传播的内在机理,包括其小世界和无尺度特性,并以电路实例说明了建模的整个过程,得出节点度数据、最大概率传播路径、聚类系数、平均路径长度等可作为诊断信息的各项模型参数。同时对电路进行了其小世界特征的小波多分辨分析以及鲁棒性和脆弱性分析。最后,将所构建的故障传播网络模型应用于实例电路的故障传播路径搜索、故障定位以及可测点优化设计,实例仿真表明了其准确性和有效性。更多还原

【Abstract】 Analog circuit fault diagnosis is always the research focus in circuit and system field.However, available data for circuit diagnosis is limited in the large scale circuits especially with package design.Even if information process with various testing signal is adopted,it’s very difficult to solve the problem of diagnosis information deficiency. The problem restricts the further development of analog circuit fault diagnosis. As we know, the fault signals propagating through the circuit are the important diagnosis information, but at the same time they can cause great arduousness and severe interference for the identification and location of true fault sources.Therefore,it is necessary to trace the fault signal and analyze the fault propagation characteristics.For this reason, this paper aims at the analysis method, simulation model and its application of fault propagation characteristics in analog circuits.At first, in view of the affinity between fault propagation characteristic and the circuit topology, this thesis verifies the small-world and scale-free feature of large-scale circuits with the case study of benchmark circuits for test,and discusses the method and implement for hierarchical division of large scale circuits.Meanwhile, under the consideration of the uncertainty of fault pattern division, cloud theory is introduced to implement the fault pattern division of circuits and the initial sample point spread. On this basis, a behavior modeling method based on data field for fault propagation is proposed. The proposed method can describe interaction of the fault patterns and fault propagation behavior.In order to analyze fault propagation characteristics of analog circuits, a Co-simulation and modeling method for fault propagation behavior is proposed with a case study to explain its procedure. The proposed method combines graph theory, fuzzy theory and probability theory with complex network. The internal mechanism of fault propagation such as small world and scale-free characteristics are found from the model parameters. The parameters such as node degree data, maximum probability propagation path, clustering coefficient, average path length and so on can be applied in fault diagnosis. Then, this paper focuses on the problems that wavelet multi-resolution analysis of the small-world feature and on the robustness and vulnerability analysis.Furthermore, this paper applies the proposed complex network model of fault propagation to searching fault propagation paths,faults location, and accessible nodes design.The case study indicates its accuracy and efficiency.更多还原