基于熵权灰色关联与改良三比值法的换流变故障演化诊断OA
Converter fault evolution diagnosis based on entropy-weighted grey relational analysis and improved three-ratio method
针对特高压换流变压器潜伏性故障诊断中,传统比值法存在的编码盲区及演化态势量化不足问题,本文以一起典型局部过热故障为例,提出了一种融合改良三比值法与熵权-灰色关联分析(EW-GRA)的多维故障诊断策略.首先,通过追踪在线监测数据趋势,有效排除传感器误报干扰;其次,应用改良的三比值法,解决 IEC 传统编码边界模糊问题,从而定性判定故障类型为高于700℃的高温过热故障.在此基础上,引入熵权法计算特征气体的客观权重系数(其中乙炔权重为 0.298),并通过灰色关联分析构建故障演化轨迹模型,量化评估故障严重程度.计算结果表明,故障关联度在关键发展期内由0.33 跃升至0.88,揭示了故障由量变累积向质变定型演化的非线性特征.最后,基于特征气体的空间浓度梯度分布锁定故障源,现场解体检查证实,该故障系紧固螺栓规格误用引发引线接头接触不良所致.该研究验证了所提诊断方法在复杂工况下具备良好的准确性和鲁棒性,可为特高压核心设备运维决策提供科学依据.
In response to the problems of coding blind spots in traditional ratio methods and the insufficient quantification of evolutionary trends in diagnosing latent faults of UHV converter transformers,this paper proposes a multi-dimensional diag-nostic strategy fusing the Modified Three-Ratio Method with Entropy-Weighted Grey Relational Analysis(EW-GRA),based on a typical local overheating case.First,online monitoring trend tracking is utilized to effectively eliminate sensor false a-larms.Second,the Modified Three-Ratio Method is applied to resolve the boundary ambiguity of traditional IEC codes,quali-tatively identifying the fault nature as high-temperature overheating exceeds 700℃.Furthermore,the Entropy Weight Meth-od is introduced to objectively calculate the weight coefficients of characteristic gases(with acetylene weight reaching 0.298),and a fault evolution trajectory model is constructed via Grey Relational Analysis to quantitatively assess fault severi-ty.The calculation results indicate that the fault correlation degree surged from 0.33 to 0.88 during the critical period,re-vealing the non-linear characteristics of the fault evolving from quantitative accumulation to qualitative determination.Final-ly,the fault source was pinpointed based on the spatial concentration gradient of characteristic gases.On-site disassembly confirmed that the fault was caused by insufficient contact at the lead connector due to the misuse of bolt specifications.This study validates the accuracy and robustness of the proposed method under complex operating conditions,providing a scientif-ic basis for operation and maintenance decisions of core UHV equipment.
郭子豪;陈斌;黄小杰
江西水利电力大学 理学院,江西 南昌 330099江西水利电力大学 理学院,江西 南昌 330099江西水利电力大学 理学院,江西 南昌 330099||江西水利电力大学 工程数学与先进计算重点实验室,江西 南昌 33099
信息技术与安全科学
换流变压器改良三比值法熵权-灰色关联分析局部过热故障诊断
converter transformermodified three-ratio methodentropy-weighted grey relational analysis(EW-GRA)local overheatingfault diagnosis
《江西水利电力大学学报》 2026 (4)
36-45,10
国家自然科学基金项目(12461014)
评论