基于小波-主成分分析的雷电过电压识别系统设计与实现

发布时间：2018-04-06 04:23

本文选题：小波变换　切入点：主成分分析　出处：《华北电力大学(北京)》2016年硕士论文

【摘要】：雷电过电压产生于电力系统外部,具有冲击电流大、冲击电压高等特点,相对于内部过电压对电力系统的危害更大。我国电力系统分布广,复杂度高,运行稳定性要求高,为了保证电网建设的经济性,针对不同雷击过电压的类型所采取的防雷措施也不同,雷电过电压相对于内部过电压防护难度更大,防护效果更加局限。因此,雷电过电压的识别可以为输电线路雷击故障的监测、检测、处理与维修提供了重要的参考价值。本文以小波分析、主成分分析和神经网络等理论为基础,针对输电线路的雷电过电压分类识别问题,深入研究了基于小波-主成分分析的数据分析和特征量提取方法,以及基于改进的神经网络的分类识别方法,并介绍了基于小波-主成分分析的雷电过电压识别系统的软件设计与实现。本论文的工作主要体现在以下三个方面：1.对雷电过电压数据的处理：针对雷电过电压高频、瞬变等特点,利用小波变换良好的时频域局部性能和多分辨率分析等特点和主成分分析法降低数据维度来提取数据特征的思想,提出了一种基于小波变换和主成分分析方法的雷电过电压分析和特征量提取方法。2.基于神经网络的分类模型的改进：针对传统BP网络对数据的分类速度慢,模型训练效率低的缺点,对BP神经网络的训练过程进行改进,提高了训练的收敛速度,并在此基础上进一步提高了分类准确率。3.系统的设计与实现：采用Java-web技术,结合MVC框架思想,实现了一个具有对雷电过电压数据的处理、分析、识别等功能的软件系统。本论文中已经验证了该分类识别模型的有效性,并且实现了基于该分类识别模型的软件系统,在接下来的研究中将继续完善该系统。
[Abstract]:Lightning overvoltage is produced from the outside of power system and has the characteristics of large impulse current and high impulse voltage. It is more harmful to power system than internal overvoltage.In order to ensure the economy of power grid construction, different lightning protection measures are adopted for different types of lightning overvoltage in order to ensure the economy of power grid construction, because of the wide distribution, high complexity and high operational stability of power system in China.Lightning overvoltage is more difficult to protect than internal overvoltage, and the protective effect is more limited.Therefore, the identification of lightning overvoltage can provide an important reference value for the monitoring, detection, processing and maintenance of lightning strike faults of transmission lines.Based on the theories of wavelet analysis, principal component analysis and neural network, this paper deeply studies the method of data analysis and feature extraction based on wavelet principal component analysis, aiming at the problem of lightning overvoltage classification and recognition of transmission lines.The classification and recognition method based on improved neural network and the software design and implementation of lightning overvoltage recognition system based on wavelet principal component analysis are introduced.The work of this paper is mainly reflected in the following three aspects: 1.Processing of lightning overvoltage data: aiming at the characteristics of lightning overvoltage high frequency, transient, etc.,Based on the advantages of wavelet transform in time-frequency domain local performance and multi-resolution analysis, and the idea of reducing the dimension of data by principal component analysis (PCA), the idea of extracting data features is presented.A method of lightning overvoltage analysis and feature extraction based on wavelet transform and principal component analysis (PCA) is proposed.The improvement of classification model based on neural network: aiming at the shortcomings of traditional BP neural network in data classification and low efficiency of model training, the training process of BP neural network is improved, and the convergence speed of training is improved.On this basis, the classification accuracy is further improved.The design and implementation of the system: a software system with the functions of processing, analyzing and recognizing lightning overvoltage data is realized by using Java-web technology and MVC framework.This paper has verified the validity of the classification and recognition model, and has implemented the software system based on the classification and recognition model, and will continue to improve the system in the following research.
【学位授予单位】：华北电力大学(北京)
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TM863

【参考文献】