高级检索

基于声纹振动特征的变压器故障诊断与状态评估方法探索

Exploration of Transformer Fault Diagnosis and Condition Assessment Methods Based on Acoustic-vibration Signatures

  • 摘要: 针对现有声纹监测中传感器配置缺乏依据、多维特征量体系不完整等问题,文章基于振动传导波长约束与绕组三相空间分布,推导传感器阵列参数化自适应布置准则,建立以频率复杂度、振动平稳性、多测点相关性、能量相似度与谐波比为维度的5维特征量诊断模型。各维度告警阈值综合参照GB/T 6451—2015、DL/T 1540—2016、DL/T 393—2010等行业标准、公开文献统计结果及工程实测数据确定,归纳典型故障模式下各特征量的差异化响应规律。该诊断模型用于某110 kV变电站2号主变3个月监测,结果表明边缘测点特征量组合与绕组端部轻度压紧力松动模式吻合,该判断已列为待检项,计划停电复核,为状态检修提供技术支撑。

     

    Abstract: To address the lack of basis for sensor configuration and the incomplete multidimensional feature system in existing acoustic-vibration monitoring, this paper derives parameterized adaptive arrangement criteria for sensor arrays based on vibration propagation wavelength constraints and the three-phase spatial distribution of windings. It establishes a five-dimensional feature diagnosis model with frequency complexity, vibration stationarity, multi-point correlation, energy similarity, and harmonic ratio as dimensions. Alarm thresholds for each dimension are determined with reference to industry standards such as GB/T 6451—2015, DL/T 1540—2016, DL/T 393—2010, statistical results from published literature, and engineering field measurement data, and the differentiated response patterns of each feature under typical fault modes are summarized. Three months of monitoring results for Donghua No. 2 main transformer show that the feature combination at edge measurement points is consistent with a slight loosening pattern of clamping force at the winding end. This judgment is listed as an item pending inspection, and outage verification is planned, providing technical support for condition-based maintenance.

     

/

返回文章
返回