| 王磊磊,张嵩阳,王枭,张光明,王广周,王东晖.基于听觉谱特征的变压器绕组状态检测研究*[J].,2022,41(2):216-224 |
| 基于听觉谱特征的变压器绕组状态检测研究* |
| Research on state detection of transformer winding loosenessbased on auditory spectrum features |
| 投稿时间：2021-04-04 修订日期：2022-02-28 |
| 中文摘要: |
| 绕组松动是变压器常见故障之一,对变压器的安全运行产生巨大威胁,对其进行精准的监测,对提高电力系统的安全稳定性具有十分重要的意义。基于声信号的变压器绕组松动检测,由于其具有无损检测和不需停运变压器等优点,成为近年来研究的热点。但声信号检测存在故障特征提前复杂和易受噪声干扰等缺陷,限制了其工程应用。本文提出了一种基于声信号听觉谱特征和SVM的变压器绕组松动识别方法。首先,将采集的声音信号进行去均值和消除趋势项的预处理,以减小信号采集环境和传感器性能对所采集信号的影响；然后,将预处理后的声音信号输入到多特征频率分析的听觉外周模型,经过中耳滤波器滤波、基底膜模型选频、外毛细胞模型放大、内毛细胞模型换能作用后,输出内毛细胞电压信号,实现多个特征频率听觉信号的提取,以此构成听觉谱,并在听觉谱基础上提取多种统计特征；最后,每种特征分别使用遗传算法优化的SVM进行识别实验,以验证提取特征的有效性。为进一步提高识别准确率,融合多种统计特征构成特征向量并进行测试,以此确定最优融合特征。研究表明,本文所建立的变压器绕组状态检测方法可以有效地应用于变压器故障诊断和监测中。 |
| 英文摘要: |
| Loose winding is one of the common faults of transformer, which pose a huge threat to the safe operation of transformers. Accurate monitoring of them is of great significance to improving the safety and stability of the power system. The detection of transformer winding looseness based on acoustic signals has become a research hotspot in recent years due to its advantages of non-destructive detection and no need to shut down the transformer. However, acoustic signal detection has defects such as complex fault features in advance and susceptibility to noise interference, which limits its engineering application. In this paper, a transformer winding looseness recognition method based on auditory spectrum and classification used support vector machine is proposed. Firstly, the collected sound signal was preprocessed by removing the mean and eliminating the trend term, so as to reduce the influence of signal acquisition environment and sensor performance on the collected signals. Then, the preprocessed signal was inputted into the auditory-periphery model with multiple characteristic frequencies for analysis. After filtering by the middle ear filter, frequency selection of basilar membrane model, amplification of the outer hair cell model and energy conversion of inner hair cell model, the inner hair cell potential signal was outputted to achieve the extraction of multiple characteristic frequency auditory signals, which constitutes the auditory spectrum, and a variety of statistical features were extracted from the auditory spectrum. Finally, each type of feature used a support vector machine optimized by genetic algorithm to perform recognition experiments to verify the effectiveness of the extracted features. For further improve the recognition accuracy, different statistical features were combined to form a feature vector and recognition experiments were performed to determine the optimal fusion feature. It demonstrates that the proposed transformer winding state detection method can be effectively applied to transformer fault diagnosis and monitoring. |
| DOI：10.11684/j.issn.1000-310X.2022.02.006 |
| 中文关键词: 变压器 绕组松动 状态检测 听觉谱特征 听觉外周模型 |
| 英文关键词: Transformer Winding Looseness State Detection Auditory Spectrum Features Auditory-periphery Model |
| 基金项目:基于声源空间识别定位技术的电抗器故障诊断方法研究 |
|
| 摘要点击次数: 812 |
| 全文下载次数: 427 |
|
查看全文
查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|
