光纤复合海底电缆故障诊断方法研究

发布时间：2018-06-24 07:43

本文选题：光电复合海缆 + 电气故障　；参考：《华北电力大学》2016年硕士论文

【摘要】：光电复合海底电缆因具有信息与电能同时传输等优点,在海岛和海上平台等工程应用中越来越广泛。为了缩短修复海缆故障所用时间,保证海缆运行通畅,减少经济损失,对海缆故障类型的诊断进行研究非常必要。本课题以某电力公司海底电缆在线监测项目的光电复合海缆为基础开展研究。本文首先综述了利用分布式光纤传感技术的海缆监测系统及故障诊断理论基础,介绍了小波包分析及神经网络的相关原理,明确了使用小波包与神经网络相结合的方法进行海缆故障类型的诊断,分析了海缆故障的几种类型及产生原因,根据多年以来海缆故障的测量经验及海缆发生故障时的传感光纤的温度和应变变化特点,模拟海缆故障数据;其次,利用小波分析算法对故障数据进行降噪处理,去除噪声干扰,还原原始信号中的有效信息,并利用小波包分解降噪后的数据,分别提取出每种故障的能量、标准差及小波包Shannon熵三种故障特征,作为模式识别的诊断依据;再次,针对电气及机械故障,分别建立了BP、PNN两种神经网络,确定了每种网络的最佳参数,将提取出的特征向量输入到两种网络中,对三种故障特征的故障区分度作了对比,并综合比较了这两种网络的诊断性能。最后,利用LabVIEW虚拟仪器,搭建了海底电缆故障诊断平台,可完成数据的读取、分析、诊断及结果显示。仿真结果表明,利用小波包分解提取出的三种特征向量均在不同程度表征出海缆故障特征。对于电气故障,提取标准差作为特征向量的诊断效果最好,对于机械故障,提取各频带能量作为特征向量的诊断效果最好。在神经网络的选择上,PNN网络的分类性能总体高于BP网络,更适合做海缆故障诊断的分类器。本文研究成果为海缆的故障诊断提供了一种经济、可靠的新途径,为海缆的在线故障监测提供理论基础,对海缆的故障诊断具有一定参考价值。
[Abstract]:Because of the advantages of simultaneous transmission of information and electric energy, optoelectronic composite submarine cables are more and more widely used in sea islands and offshore platforms. In order to shorten the time of repairing submarine cable fault, ensure the smooth operation of submarine cable and reduce economic loss, it is very necessary to study the diagnosis of submarine cable fault type. The research is based on the optoelectronic composite submarine cable of a submarine cable online monitoring project of a power company. In this paper, the monitoring system and fault diagnosis theory of submarine cable based on distributed optical fiber sensing technology are reviewed, and the related principles of wavelet packet analysis and neural network are introduced. In this paper, the method of combining wavelet packet with neural network is used to diagnose the fault types of submarine cables, and several types of faults and their causes are analyzed. According to the measurement experience of submarine cable fault over the years and the temperature and strain characteristics of sensing fiber when submarine cable fault occurs, the fault data of submarine cable are simulated. Secondly, the wavelet analysis algorithm is used to reduce the noise of the fault data. Removing the noise interference, reducing the effective information in the original signal, and using wavelet packet decomposing the de-noised data, the energy, standard deviation and Shannon entropy of each fault are extracted respectively, which can be used as the diagnostic basis for pattern recognition. Thirdly, for electrical and mechanical faults, two kinds of neural networks are established, and the optimal parameters of each network are determined. The extracted feature vectors are input into the two networks, and the fault classification of the three fault features is compared. The diagnostic performance of the two networks is compared. Finally, using LabVIEW virtual instrument, a submarine cable fault diagnosis platform is built, which can read, analyze, diagnose and display the data. The simulation results show that the three eigenvectors extracted by wavelet packet decomposition can represent the fault characteristics of submarine cable to varying degrees. For electrical faults, the diagnostic effect of extracting standard deviation as eigenvector is the best, and for mechanical fault, the best is to extract the energy of each frequency band as eigenvector. In the selection of neural network, the classification performance of PNN is higher than that of BP neural network, so it is more suitable to be used as classifier for submarine cable fault diagnosis. The research results in this paper provide a new economic and reliable way for the fault diagnosis of submarine cables, and provide a theoretical basis for on-line fault monitoring of submarine cables, and have a certain reference value for the fault diagnosis of submarine cables.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：P756.1

【参考文献】