基于异常振动分析的GIS机械故障诊断技术研究

发布时间：2018-03-13 12:05

本文选题：GIS　切入点：机械故障诊断　出处：《山东大学》2017年硕士论文　论文类型：学位论文

【摘要】：随着我国科学技术的快速发展,对于现代电力的要求越来越高。近年来,伴随着电压等级的不断提高,对电能质量的可靠性也提出了更高的要求。SF6气体绝缘金属封闭式组合电器(GIS)由于具有占地面积小、可靠性高等特点,成为电力系统的重要组成部分。但近年来,随着国民用电的进一步提高,电网故障事故频发,而GIS作为重要环节,一旦出现故障就可能造成大规模的停电事故,致使电网损失惨重。所以,保证GIS的安全运行利国利民。本文致力于研究GIS两种常见的机械故障,最关键的是从振动信号中提取对应故障的特征信息。对变分模态分解(VMD)算法的基本原理进行详细说明,针对它的参数不易确定的特性,提出了运用混沌布谷鸟算法优化VMD参数的方法。并通过测试函数对标准布谷鸟优化算法和混沌布谷鸟优化算法进行仿真对比,验证了混沌布谷鸟优化算法无论是收敛性还是搜索速度上均优于标准布谷鸟优化算法。以实验室仿真GIS试验段作为研究对象,构造螺丝松动和金属微粒振动两种常见GIS故障,提出了一种基于混沌布谷鸟算法优化VMD参数的GIS机械故障诊断方法,首先运用混沌布谷鸟算法优化VMD参数;然后对GIS正常振动信号和故障信号进行VMD分解,提取不同故障类型信号的特征向量;最后通过线性递减权重PSO优化的K-means聚类算法对训练样本特征向量集进行聚类,得到不同的聚类中心,再利用最小欧氏距离原则对测试样本特征向量集进行识别,实现GIS机械故障的诊断。基于理论研究与仿真分析结果,本文以VMD算法为核心设计了一套软硬件完备的基于异常振动分析的GIS机械故障在线监测系统。其硬件部分主要包含数据采集模块、信号输出模块以及信号分析模块,数据采集模块主要负责采集GIS振动信号;信号输出模块以数据采集卡为主体,实现数据的A/D转换功能;信号分析模块则是数据经采集卡导入微机处理,最终实现信号特征提取。软件部分则以VMD算法为核心,实现GIS常见故障的特征提取,本文还添加了数据存储和数据检索模块,极大地丰富了软件的功能及其扩展性。本文利用现场实际运行的GIS作为振动源对开发的基于异常振动分析的GIS机械故障诊断系统进行整体测试。检测发现:当GIS出现异常的机械振动时,诊断系统VMD熵值往往异常正常值,并且变化明显,弥补了传统超高频和超声波无法有效检测低频机械信号的不足,验证了本文构建的基于异常振动分析的GIS机械故障系统的有效性。
[Abstract]:With the rapid development of science and technology in China, the demand for modern power is becoming higher and higher. Higher requirements for reliability of power quality. SF6 gas insulated metal enclosed composite electrical apparatus (GISs) has become an important part of power system because of its small area and high reliability. With the further improvement of national power consumption, power grid failures occur frequently, and GIS, as an important link, may cause large-scale power outages once they occur. This paper is devoted to the study of two common mechanical faults of GIS, the most important of which is to extract the characteristic information of the corresponding faults from the vibration signal. The basic principle of the variational mode decomposition (VMD) algorithm is explained in detail. In view of the difficulty of determining its parameters, a method of optimizing VMD parameters by using chaotic cuckoo algorithm is proposed, and the standard cuckoo optimization algorithm and chaotic cuckoo optimization algorithm are simulated and compared by testing function. It is verified that the chaotic cuckoo optimization algorithm is superior to the standard Cuckoo optimization algorithm in terms of convergence and search speed. Taking the laboratory simulation GIS test section as the research object, the construction screw loosening and metal particle vibration are two common GIS faults. This paper presents a GIS mechanical fault diagnosis method based on chaotic cuckoo algorithm to optimize VMD parameters. Firstly, chaotic cuckoo algorithm is used to optimize VMD parameters, and then GIS normal vibration signal and fault signal are decomposed by VMD. The feature vectors of different fault types are extracted. Finally, different clustering centers are obtained by using K-means clustering algorithm optimized by linear decrement weight PSO to cluster the feature vector sets of training samples. Then using the principle of minimum Euclidean distance to identify the characteristic vector set of test samples to realize the diagnosis of GIS mechanical fault, based on the theoretical research and simulation analysis results, Based on the VMD algorithm, this paper designs a complete GIS mechanical fault on-line monitoring system based on abnormal vibration analysis, which includes data acquisition module, signal output module and signal analysis module. The data acquisition module is mainly responsible for collecting the GIS vibration signal, the signal output module takes the data acquisition card as the main body, realizes the A- / D conversion function of the data, the signal analysis module is the data through the acquisition card to import the microcomputer processing, Finally, the signal feature extraction is realized. The software part uses VMD algorithm as the core to realize the feature extraction of common faults of GIS. This paper also adds the module of data storage and data retrieval. The function and extensibility of the software are greatly enriched. In this paper, the GIS mechanical fault diagnosis system based on abnormal vibration analysis is tested by using the field running GIS as the vibration source. When abnormal mechanical vibration occurs, The VMD entropy value of diagnostic system is often abnormal, and the change is obvious, which makes up for the deficiency that the traditional UHF and ultrasonic can not detect the low frequency mechanical signal effectively. The validity of the GIS mechanical fault system based on abnormal vibration analysis is verified.
【学位授予单位】：山东大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM595

【相似文献】