电缆终端局部放电信号去噪方法研究

发布时间：2018-10-30 18:18

【摘要】：局部放电检测是电气设备故障诊断最有效的方法之一,被广泛应用于电缆附件故障诊断。但是,由于背景噪声的存在,尤其是白噪声的存在使得局部放电在线检测的灵敏度大大降低。白噪声的频率成分遍布整个频域,时域波形也遍布整个时域,这使得去除局放信号中的白噪声干扰成为局放信号去噪技术中的一个难点和热点研究课题。基于以上论述,本文以电缆终端局部放电检测为基础提出了两种局放信号去噪方法分别用以去除高信噪比和低信噪比下的白噪声干扰。主要研究内容如下:(1)分析了电缆终端发生故障的主要原因;设计制作了 4种电缆终端典型缺陷模型作为本文试验部分的试验对象开展局部放电试验;采集了 4种电缆终端典型缺陷的局放信号作为本文所提去噪方法的试验验证。(2)提出了一种基于峭度和时域能量的局放信号去噪方法,用以解决高信噪比下现有局放信号去噪方法计算速率慢、去噪后脉冲边沿不清晰的问题。首先通过计算染噪局放信号峭度实现局放脉冲的定位,进而确定局放脉冲峰值在整个数据窗中的位置以减小后续脉冲提取的计算量;其次通过分析白噪声与局放脉冲信号的时域能量差异,确定时域能量阈值及对应的时窗长度;最后以脉冲峰值为中心采用时域能量搜索的方法向两边搜索脉冲的边沿实现局部放电脉冲边沿检测及确定,进而实现局部放电脉冲提取。(3)提出了一种局放信号自适应稀疏分解去噪方法,同时构造了与局放信号对应的匹配原子库,用以提高局放信号去噪过程中MP算法的计算效率和去噪效果。基于信号快速谱峭度和S变换,迅速获取局放信号时频特性,其中包括中心频率、带宽、局放发生起始点和熄灭点的大概位置;利用局放信号的时频特性优化局放信号匹配原子库的时频参数,在对染噪局放信号进行MP计算时自适应选择少量原子进行寻优匹配;用各次迭代所得最佳匹配原子对原始局放脉冲信号进行稀疏表示,达到局放信号去噪目的。仿真及试验表明,本文所提出的两种局放信号去噪方法有效的解决了现有局放信号去噪方法去噪过程中所存在的去噪程序繁琐、计算速率慢、去噪不彻底、去噪后波形畸变问题。
[Abstract]:Partial discharge detection is one of the most effective methods for fault diagnosis of electrical equipment and is widely used in cable accessory fault diagnosis. However, because of the background noise, especially the white noise, the sensitivity of PD on-line detection is greatly reduced. The frequency components of white noise are all over the frequency domain, and the waveform of time domain is all over the time domain, which makes the removal of white noise interference from partial discharge signal become a difficult and hot research topic in the denoising technology of partial discharge signal. Based on the above discussion, based on the partial discharge detection of cable terminal, two kinds of partial discharge signal denoising methods are proposed to remove white noise interference at high signal-to-noise ratio and low signal-to-noise ratio respectively. The main research contents are as follows: (1) the main causes of cable terminal failure are analyzed, and four typical defect models of cable terminal are designed and made as the test object of this paper to carry out partial discharge test. Four kinds of partial discharge signals with typical defects of cable terminals are collected as experimental verification of the denoising methods proposed in this paper. (2) A denoising method for partial discharge signals based on kurtosis and time domain energy is proposed. It is used to solve the problem of slow calculation speed and unclear pulse edge of existing partial discharge signal denoising methods under high SNR. Firstly, the location of partial discharge pulse is realized by calculating the kurtosis of local discharge signal, and then the position of peak value of partial discharge pulse in the whole data window is determined to reduce the computation amount of subsequent pulse extraction. Secondly, by analyzing the time domain energy difference between white noise and partial discharge pulse signal, the time domain energy threshold and the corresponding time window length are determined. Finally, using the time-domain energy search method to search the edge of the pulse on both sides, the detection and determination of the partial discharge pulse edge can be realized by taking the peak value of the pulse as the center. Then partial discharge pulse extraction is realized. (3) an adaptive sparse decomposition de-noising method for partial discharge signal is proposed, and a matching atom library corresponding to partial discharge signal is constructed. In order to improve the computational efficiency and denoising effect of MP algorithm in the process of partial discharge signal denoising. Based on the fast spectral kurtosis and S-transform, the time-frequency characteristics of PD signal are obtained, including the center frequency, bandwidth, the starting point of PD and the approximate position of extinguishing point. The time-frequency parameters of the partial discharge signal matching atom library are optimized by using the time-frequency characteristic of partial discharge signal, and a small number of atoms are adaptively selected for optimal matching when MP calculation is carried out on the noisy partial discharge signal. The original PD pulse signal is represented sparsely by the best matching atoms obtained from each iteration to achieve the purpose of de-noising the PD signal. The simulation and experiments show that the two methods proposed in this paper can effectively solve the problem that the existing partial discharge signal denoising methods are complicated in the process of de-noising, slow calculation rate and incomplete de-noising. Waveform distortion after denoising.
【学位授予单位】：西南石油大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TM855

【参考文献】