基于小波分析和FastICA算法在电磁兼容中的研究

发布时间：2018-04-26 01:07

本文选题：电磁兼容 + 电磁干扰　；参考：《南京师范大学》2015年硕士论文

【摘要】：随着高速功率开关器件在电气电子产品中的普遍使用,电力电子装置的大量传导性电磁干扰(EMI)问题在现实中不断出现并未得到很好的解决,已成为现代电力电子技术进步的一个重要约束,使得人们开始对电磁兼容问题重新重视。因此对电力电子设备的传导性电磁干扰的诊断和产生机理的研究具有十分重要的意义。本文根据国内外对电磁兼容的动态和研究现状,提出了基于FastICA算法和小波分析方法在电磁兼容中的应用。首先对独立分量分析和小波分析理论做出详细的阐述,并且通过Matlab仿真验证该方法的有效性。以PI (Perfomance Index)性能指标来衡量FastICA算法的性能,并且采用Pearson(皮尔逊)相关系数的相关度来说明分离或提取结果的有效性。首先以开关电源为例,分析传导电磁干扰噪声的产生机理和传播途径,通过人工电源网络(LISN)将电源线上的总噪声提取出来,并利用硬分离网络分离出共模噪声信号和差模噪声信号,通过皮尔逊相关性对比分析基于硬件模态分离网络的模态分离和基于FastICA算法和小波分析方法结合的模态分离。结果表明该方法对传导电磁干扰的模态分离是有效的,从而对电磁兼容中模态分离过程进行了简化,大大减少了对硬件电路的依赖,降低了成本和复杂性。其次搭建了噪声源提取模型,使用人工电源网络提取出电源线上的总噪声。文中通过实验不仅验证了基于FastICA算法提取噪声源的有效性,还通过皮尔逊相关性对比分析经过小波分析后再应用FastICA算法对总噪声进行源信号提取的优势。结果表明该方法对传导电磁干扰的噪声源提取的有效性,精确、快速地解决了以电子器件为核心的传导噪声源识别方面的问题,简洁高效且系统化,更加准确的判断产生传导噪声信号的电子器件,能够更有针对性的减少电磁干扰。通过这种方法,不仅满足技术指标,更能最大程度地降低对硬件的依赖,节省经济成本,实现技术和经济的一体化。
[Abstract]:With the widespread use of high-speed power switch devices in electrical and electronic products, the problem of a large number of conductive electromagnetic interference (EMI) in power electronic devices has not been solved well in reality. It has become an important constraint for the progress of modern power electronics technology, which makes people begin to pay more attention to the problem of electromagnetic compatibility (EMC). Therefore, it is of great significance to study the diagnosis and generation mechanism of conductive electromagnetic interference (EMI) in power electronic equipment. According to the dynamic and research status of EMC at home and abroad, this paper presents the application of EMC based on FastICA algorithm and wavelet analysis method. The theory of independent component analysis and wavelet analysis is described in detail, and the effectiveness of the method is verified by Matlab simulation. The performance of FastICA algorithm is evaluated by Pi Perfomance Index, and the correlation degree of Pearson correlation coefficient is used to illustrate the effectiveness of the separation or extraction results. Firstly, taking switching power supply as an example, this paper analyzes the generation mechanism and propagation way of conducting electromagnetic interference noise, and extracts the total noise from power supply line by artificial power supply network (LISN). The common mode noise signal and differential mode noise signal are separated by hard separation network. The modal separation based on hardware modal separation network and the combination of FastICA algorithm and wavelet analysis method are analyzed by Pearson correlation comparison. The results show that the proposed method is effective for the mode separation of electromagnetic interference, which simplifies the mode separation process in EMC, greatly reduces the dependence on hardware circuits, and reduces the cost and complexity. Secondly, the noise source extraction model is built, and the total noise on the power line is extracted by artificial power network. The experiment not only verifies the validity of extracting noise source based on FastICA algorithm, but also uses FastICA algorithm to extract the total noise signal after wavelet analysis through Pearson correlation comparison analysis. The results show that the method is effective and efficient in the noise source extraction of conductive electromagnetic interference. It is accurate and fast to solve the problem of the identification of conductive noise sources with electronic devices as the core. It is simple, efficient and systematic. It is more accurate to judge the electronic devices which produce the conduction noise signal, which can reduce the electromagnetic interference more pertinently. This method can not only satisfy the technical index, but also reduce the dependence on hardware, save the economic cost and realize the integration of technology and economy.
【学位授予单位】：南京师范大学
【学位级别】：硕士
【学位授予年份】：2015
【分类号】：TN03

【参考文献】