基于IDDT的CMOS数字电路故障诊断技术研究-检测技术与自动化装置专业论文.docxVIP

摘 要 传统的基于电压测试的测试方法己得到了广泛的应用，但这种方法仍然无法有效 地检测某些故障。作为电压测试方法的补充，电流测试方法能够提高故障覆盖率和产 品的可靠性。瞬态电流提供了一个观测电路内部开关性能的新窗口，瞬态电流测试方 法为进一步提高故障覆盖率提供了可能。 本文分析CMOS电路主要电流组成成份的物理意义和瞬态电流CIDDT )的特性， 从瞬态电流CIDDT )中可以提取比静态电流CIDDQ)具有更丰富的故障信息。探讨了 小波变换和神经网络的理论及应用于数字电路故障诊断的优势。针对数字电路中的桥 接故障、电阻性开路故障，研究了瞬态电流CIDDT )测试与小波-神经网络相结合的数 字电路故障诊断技术。利用小波变换技术对瞬态电源电流进行故障特征提取，BP神经 网络用来进行故障类型的识别。为了降低故障样本的维数，采用主成分分析和归一化 数据预处理，提高了神经网络的训练效率。衷方法克服了传统故障字典方法的局限性， 实现了数字电路快速和准确的故障定位。 通过对数字电路的仿真结果和数据分析表明，本文所采用的小波和神经网络故障 诊断算法具有快速、准确的特点，达到了较理想的故障诊断效果。 关键词:瞬态电流测试; 故障诊断; BP神经网络; 小波分析; 电阻性开路故障. 万方数据 – I – Abstract The traditional testing method based on voltage has been widely used in practice. But there are still some defects in circuit that can not be detected by traditional testing method. As a supplement to this method, current testing method can increase the fault coverage and make the higher reliability of product. The transient current has offered a new window of observing switch performance of circuit. Transient current testing (IDDT ) method makes possible further increasing the fault coverage. Firstly, An arithmetic model of IDDT ’s average value of a single period is studied by means of analyzing the components of the current in CMOS. By study on the characteristics of the IDDT , we can prove that IDDT does have more information of defect than IDDQ.Secondly,The paper investigates the theory of wavelet analysis and neural newtokr，and its feasibility applying to fault diagnosis is analyzed. later, a novel integrated intelligent fault diagnosis method based on the BP neural network and the IDDT testing of digital circuit has been researched for some hard faults, such as bridge fault and resistive-open fault, considering good properties of Wavelet analysis, fault feature of the sampled signal normalization and principal component analysis, which reduced the date dimension and shorten the training time. This method overcomes the location fault dictionary method and reaches a high and accurate fault d