基于人工免疫算法的模拟电路故障诊断

【作者】 梅聪；

【导师】 孙建红；

【作者基本信息】 南京理工大学 ， 电路与系统， 2010， 硕士

【摘要】 模拟电路故障诊断一直以来都是十分必要和有意义的,目前已成为热门的研究课题。现代电子技术和计算机技术的迅速发展促进了片上系统和混合集成电路的大量涌现,这也对模拟电路的测试和故障诊断提出了更高的要求。传统模拟电路故障诊断方法受到元件误差影响较大,其诊断过程需求解大量的非线性方程,计算量大,诊断效果不是很理想。而现代智能技术为模拟电路故障诊断提供了一条有效的途径。人工免疫系统具有非常适合模拟电路故障诊断的特性,它无需建立诊断对象的精确数学模型,特别是其自组织、自学习和记忆等的能力,使其在故障诊断中具有很好的应用前景。本文首先针对故障样本信息的获取进行了研究,研究了模拟电路故障特征向量的提取方法,包括基于小波变换和有效点采样的特征向量提取方法；然后深入地分析了人工免疫算法,总结出免疫算法的一般框架,在免疫算法的具体方法设计上进行了探讨；最后研究了基于克隆选择算法和人工免疫网络的模拟故障诊断方法。针对克隆选择算法本身的一些缺陷,如就收敛速度慢,容易陷入局部收敛的缺点,提出了自适应的变异算子、交叉算子和小生境技术改进的方法。该方法首先将每个抗体克隆后的集合作为小生境的子种群,分别对每个子种群利用自适应克隆变异和交叉操作进行全局和局部搜索的调节,对于提高电路故障诊断的速度有很大帮助,诊断效率也有一定的提高。人工免疫网络的学习算法就是利用免疫网络的激励和抑制、克隆选择等自适应学习机制来产生一个能够对应和识别抗原的记忆抗体种群。人工免疫网络为故障样本的约减和聚类提供一种有效的方法。针对模拟电路故障诊断的特点改进了人工免疫网络,将自适应变异算子进一步改进应用于模拟电路故障诊断中,通过仿真诊断结果显示该算法具有很好的样本约减能力和很高的故障诊断率,证明了该算法的有效性和可行性。更多还原

【Abstract】 The theory of analog circuit fault diagnosis is very important and significant,and now it has become a hot research topic.The rapid progress in morden electronic and computer technology promotes the advent of system-on-chip and mixed-signal integrated circuits,which present higher and newer circuit test request.Due to the impact of tolerance components,manys traditional analog circuit fault diagnosis methods are not very efficiency.Meanwhile,they need to calculate a large number of nonlinear equations,so the workload is excessive and the results are not satisfactory.Mordern intelligent technologies provide a effective way for circuit fault diagnosis.Artificial immune system has the features which are suitable for analog circuit fault diagnosis.It does not need to establish the precise mathematic model. In the fault diagnosis it has a good prospect because of Its self-organization, self-learning and memory capacities.In order to solve the problem how to access the circuit fault feature, the the article studies the fault feature methods,including effective sampling points extraction and wavelet analysis. The article researches the artificial immune system deeply,sums up the general framework of the immune algorithm and discusses the design of the specific immune algorithm.The paper investigates the analog circuit fault diagnosis method based on clonal selection algorithm and artificial immune network.Clone selection algorithm has a number of shortcomings such as slow convergence,easily caught in the local minimum value. The paper uses adaptive mutation、crossover and Niche to overcome the shortcomings.The method takes each cloned population of each antibody as the Niche sub-population and implements crossover and adaptive mutation on the Niche sub-population respectively.The simulation shows that the method can increase the diagnosis speed and diagnostic accuracy.The learning algorithm of artificial immune network uses the adaptive learning mechanism such as promotion and inhibtion of the immune network and clone selection,to generate an memory antibody population which can identify the antigen. The algorithm provides the methods to reduce and cluster the fault samples, further Improved adaptive mutation is applied to analog circuit fault diagnosis. The diagnostic results show that the algorithm has a good ability to reduce the diagnostic samples and a high rate of fault diagnosis,and proves the effectiveness and feasibility of the algorithm.更多还原