脉冲神经膜系统在电力系统故障诊断中的应用研究

发布时间：2018-10-29 14:29

【摘要】：随着我国经济的快速发展,各行各业对电力的需求不断增加,电力系统的稳定运行已成为关系国计民生的大事。但由于电力系统规模庞大,结构复杂,且长时间暴露于恶劣的自然条件下,故障的发生难以避免。同时,地铁牵引供电系统作为地铁的能源系统,其安全可靠运行对地铁稳定运行有着重要意义。但近年来,由于地铁牵引供电系统故障而造成列车停运、晚点的事故时有发生。因此,当故障发生时,需要迅速诊断并隔离故障。但在实际运行中,传统故障诊断方法并没有较好解决电力系统故障诊断问题,错判、误判仍时有发生。因此,一方面是对原有故障诊断方法进行改进,另一方面是探索新的故障诊断方法。脉冲神经膜系统是一种新颖的具有分布式并行计算能力的计算模型,且具有较好的动态特性。基于此,许多研究将其应用于解决实际问题。本文则将脉冲神经膜系统应用于电力系统故障诊断与地铁牵引供电系统故障诊断问题中,主要包括以下3点工作:(1)给出基于波形相似度的输电线路故障可信度。利用小波变换理论分析了输电线路出现故障时,其波形信号出现的幅值与谐波变化,并采用波形相关系数来反映线路故障时其幅值与谐波的变化程度。同时,为了验证波形相似度故障可信度能否有效准确反映输电线路故障,利用PSCAD建立模型,仿真180种不同的故障情况,验证其有效性;(2)将脉冲神经膜系统应用于故障选相中。建立基于脉冲神经膜系统的故障选相模型,引入6种故障选相特征值,并分别给出其计算方法。给出基于故障选相模型的故障选相推理算法,实现故障选相。利用PSCAD建立模型,仿真450种不同的故障类型,验证所提选相方法有效性;(3)将脉冲神经膜系统应用于地铁牵引供电系统故障诊断中。给出基于网络拓扑分析法的故障区域确定方法,确定疑似故障元件。对疑似故障元件分别建立基于脉冲神经膜系统的故障诊断模型,这些故障诊断模型分别运行计算得到各个疑似故障元件的故障可信度,从而确定故障元件。
[Abstract]:With the rapid development of economy in our country, the demand for electricity in various industries is increasing, and the stable operation of power system has become a major event related to the national economy and the people's livelihood. However, due to the large scale of power system, complex structure and long time exposure to harsh natural conditions, it is difficult to avoid the fault. At the same time, as the energy system of subway, its safe and reliable operation is of great significance to the steady operation of subway. But in recent years, due to the subway traction power supply system failure caused by train shutdown, late accidents have occurred from time to time. Therefore, when the fault occurs, it is necessary to diagnose and isolate the fault quickly. However, in practice, the traditional fault diagnosis method has not solved the problem of power system fault diagnosis. Misjudgment and misjudgment still occur from time to time. Therefore, on the one hand, the original fault diagnosis method is improved, on the other hand, a new fault diagnosis method is explored. Pulse neural membrane system is a novel computing model with distributed parallel computing capability and has good dynamic characteristics. Based on this, many researches apply it to solve practical problems. In this paper, the pulse neural membrane system is applied to power system fault diagnosis and subway traction power supply system fault diagnosis, mainly including the following three points: (1) the reliability of transmission line fault based on waveform similarity is given. The wavelet transform theory is used to analyze the amplitude and harmonic variation of the waveform signal when the transmission line fails, and the waveform correlation coefficient is used to reflect the amplitude and harmonic variation degree of the transmission line fault. At the same time, in order to verify whether the fault credibility of waveform similarity can reflect the transmission line fault effectively and accurately, a model is established by using PSCAD to simulate 180 different fault cases, and the validity of the model is verified. (2) the pulse nerve membrane system is applied to fault phase selection. A fault phase selection model based on pulsed neural membrane system is established. Six eigenvalues of fault phase selection are introduced and their calculation methods are given. A fault phase selection reasoning algorithm based on fault phase selection model is presented to realize fault phase selection. The PSCAD model is used to simulate 450 different fault types to verify the effectiveness of the proposed phase selection method. (3) the pulse neural membrane system is applied to the fault diagnosis of metro traction power supply system. A method of fault area determination based on network topology analysis is presented to determine the suspected fault elements. The fault diagnosis models based on the pulse neural membrane system are established for the suspected fault elements. The fault reliability of each suspected fault element is calculated by running these fault diagnosis models, and the fault components are determined.
【学位授予单位】：西南交通大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM711

【相似文献】