基于混沌神经网络的模拟电路故障诊断方法研究

发布时间：2019-02-15 05:20

【摘要】：随着电子技术的进步,模拟电路向着大规模、集成化方向发展,电路元件发生故障的几率大大增加,个别元件发生故障往往会影响到整个系统的正常工作。科技的发展对模拟电路故障的测试和诊断提出了更加严格的要求,使得对模拟电路故障领域的研究显得尤为重要,一些传统的故障诊断方法不再适用于现在的电路。随着以神经网络为代表的人工智能技术以及小波理论、混沌、粒子群等新方法不断融合到故障诊断领域中,这些新方法为模拟电路故障诊断开辟了新的思路,使模拟电路故障诊断方法的研究趋于多样化与实用化。进行高效的故障诊断方法研究是发展的趋势。本论文探讨了模拟电路故障诊断,研究了基于神经网络及混沌神经网络的模拟电路故障诊断方法,主要内容由以下几部分构成:(1)详细阐述了神经网络理论,对神经网络的灵活应用特性及有效分类性能进行了研究,并应用小波神经网络进行模拟电路故障诊断。(2)在故障特征提取环节中,提出一种小波分解与模糊聚类相结合进行预处理的故障特征提取方法,滤除冗余及噪声分量,有效提取故障特征,实现降低故障信息维数,优化神经网络的输入量。(3)提出一种基于混沌神经网络的模拟电路故障诊断方法,把混沌的遍历性、随机性和对初值的敏感性等特点与神经网络的学习、映射能力相结合,使神经网络具有混沌特性,改善了神经网络的收敛速度,提高了网络的学习、泛化能力,提高故障诊断效率。(4)为了验证文中所提方法的正确性,使用ORCAD10.5、MATLAB软件对实际电路进行编程仿真;仿真结果表明,本文所提出的方法能够进一步提高模拟电路故障诊断的效率,对软故障具有很好的故障定位能力。
[Abstract]:With the development of electronic technology, analog circuits are developing towards large scale and integration. The probability of circuit components failure is greatly increased, and the failure of individual components will often affect the normal operation of the whole system. With the development of science and technology, more strict requirements are put forward for the fault testing and diagnosis of analog circuits, which makes the research on the fault field of analog circuits more important. Some traditional fault diagnosis methods are no longer suitable for the current circuits. With the artificial intelligence technology represented by neural network and wavelet theory, chaos, particle swarm optimization and other new methods merging into the field of fault diagnosis, these new methods open a new way for analog circuit fault diagnosis. The research on fault diagnosis of analog circuits tends to be diversified and practical. It is a developing trend to study the efficient fault diagnosis method. This paper discusses the fault diagnosis of analog circuit, and studies the fault diagnosis method of analog circuit based on neural network and chaotic neural network. The main contents are as follows: (1) the theory of neural network is expounded in detail. The flexible application characteristic and effective classification performance of neural network are studied, and the fault diagnosis of analog circuit is carried out by using wavelet neural network. (2) in fault feature extraction, A fault feature extraction method combining wavelet decomposition and fuzzy clustering is proposed to filter out redundant and noise components, extract fault features effectively and reduce the dimension of fault information. (3) an analog circuit fault diagnosis method based on chaotic neural network is proposed. The characteristics of chaos ergodicity, randomness and sensitivity to initial values are combined with the learning and mapping ability of neural network. The neural network has chaotic characteristics, improves the convergence speed of neural network, improves the learning and generalization ability of neural network, and improves the efficiency of fault diagnosis. (4) in order to verify the correctness of the proposed method, ORCAD10.5, is used. The actual circuit is simulated by MATLAB software. Simulation results show that the proposed method can further improve the efficiency of analog circuit fault diagnosis and has a good fault location ability for soft faults.
【学位授予单位】：湖南师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN710;TP183

【共引文献】