基于RPROP神经网络的输电线路故障分析研究

发布时间：2019-06-11 05:20

【摘要】：随着国民经济与电力网络的发展,电力供应的可靠性越来越受到重视,输电线路作为电力网络的重要组成部分,其运行的可靠性直接影响着整个电力网络。输电线路大多数直接暴露在自然环境中,因此输电线路也是电力系统中最易发生故障的部分。输电线路发生故障时,系统对故障相的准确识别将直接决定着继电保护装置动作的正确性。据此,本文提出了基于RPROP神经网络及小波分析理论的输电线路故障类型识别算法,主要研究基于RPROP神经网络与小波分析理论所构成的故障诊断网络在输电线路故障诊断和继电保护方面的可行性和有效性,同时提高故障类型识别的准确性和快速性。本文系统地阐述了小波分析与RPROP神经网络的相关理论,包括输电线路故障识别的研究现状、神经网络在电力系统中的研究现状和小波变换的发展应用现状等。本文采用基于小波变换的暂态电压高频分量的能量比值作为特征量与RPROP神经网络相结合的方法实现输电线路故障类型的识别。本文首先对输电线路各故障类型特征进行了分析,并根据其特征利用小波变换对故障暂态电压进行分解,并计算提取出暂态电压被分解后各频带的系数能量比。其次,结合本文的研究内容构建了三层的RPROP故障诊断网络,即输入层、隐含层和输出层各为一层,并将利用小波分析提取的故障暂态电压各频带系数能量的比值作为故障诊断网络的输入量,网络的输出量为各短路故障类型所对应的四位二进制编码。最后,利用故障数据样本对网络进行训练与测试,并对部分故障样本进行仿真验证。根据仿真验证的实验结果,确定了 RPROP网络在输电线路故障诊断中的可行性和有效。同时,还将RPROP网络的诊断结果与传统的BP网络诊断结果进行了对比分析,从而得出RPROP网络算法在输电线路故障诊断中的诊断效果优于传统的BP网络算法的诊断效果的结论。
[Abstract]:With the development of national economy and power network, more and more attention has been paid to the reliability of power supply. As an important part of power network, the reliability of transmission line operation directly affects the whole power network. Most of the transmission lines are directly exposed to the natural environment, so the transmission lines are also the most prone to failure in the power system. When the transmission line fails, the accurate identification of the fault phase will directly determine the correctness of the relay protection device. Based on this, a fault type recognition algorithm for transmission lines based on RPROP neural network and wavelet analysis theory is proposed in this paper. This paper mainly studies the feasibility and effectiveness of the fault diagnosis network based on RPROP neural network and wavelet analysis theory in transmission line fault diagnosis and relay protection, and improves the accuracy and rapidity of fault type identification. In this paper, the related theories of wavelet analysis and RPROP neural network are systematically described, including the research status of transmission line fault identification, the research status of neural network in power system and the development and application of wavelet transform. In this paper, the energy ratio of high frequency components of transient voltage based on wavelet transform is combined with RPROP neural network to identify the fault types of transmission lines. In this paper, the characteristics of each fault type of transmission line are analyzed, and the fault transient voltage is decomposed by wavelet transform according to its characteristics, and the coefficient energy ratio of each frequency band after the transient voltage is decomposed is calculated and extracted. Secondly, combined with the research content of this paper, a three-layer RPROP fault diagnosis network is constructed, that is, the input layer, the hidden layer and the output layer are each one layer. The ratio of frequency band coefficient energy of fault transient voltage extracted by wavelet analysis is used as the input of fault diagnosis network, and the output of the network is the four-bit binary coding corresponding to each short-circuit fault type. Finally, the fault data samples are used to train and test the network, and some fault samples are simulated and verified. According to the experimental results verified by simulation, the feasibility and effectiveness of RPROP network in transmission line fault diagnosis are determined. At the same time, the diagnosis results of RPROP network are compared with those of traditional BP network, and it is concluded that the diagnosis effect of RPROP network algorithm in transmission line fault diagnosis is better than that of traditional BP network algorithm.
【学位授予单位】：安徽理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM755

【相似文献】