【作者】 李颖；

【导师】 曹辰； 李辉；

【作者基本信息】 沈阳工业大学 ， 电气工程， 2025， 硕士

【摘要】 在电力系统中,变压器的运行状态对电网稳定运行具有决定性影响。变压器在长时间的运行过程中,不可避免地会受到各种干扰和冲击,进而导致其出现故障,其中绕组故障所占的比例较大。尽管振动分析法可以实现对变压器绕组振动信号的辨识,但其对设备器身及油箱振动特性的探究仍显不足。因此,本文基于220k V油浸式电力变压器,采用仿真分析、试验验证与信号特征提取相结合的方法,对变压器振动特性及其绕组在线辨识展开深入研究。首先系统研究了变压器铁芯和绕组振动的产生机理及其传递过程,重点探讨了绝缘油对油箱振动的影响。将油介质中机械振动的传递等效为声波的传递,同时将油箱简化为无限大单板结构,充分考虑声波所产生的透射波与反射波,分别建立了振动波在绝缘油中传递的声学数学模型和油箱振动的数学模型。利用有限元仿真法,建立变压器三维仿真模型。在分析过程中,通过设置电场、磁场、压力声学以及固体力学模块,实现对变压器在不同工况下的多物理场耦合仿真分析。具体来说,针对变压器空载和负载两种运行工况,重点计算铁芯和绕组的磁场分布规律及其振动特征。在此基础上,分别对空载、负载以及带载工况下油箱表面的振动分布特征进行计算。结果表明:额定负载条件下,变压器高压侧电流幅值为683.8A,与设计值相比误差为7.0%,低压侧电流幅值为2083.3A,与设计值相比误差为2.2%,绕组漏磁为0.403T。额定空载条件下,变压器高压侧电压幅值为316.5k V,与设计值相比误差为1.7%,低压侧电压幅值为99.8k V,与设计值相比误差为2.3%,铁芯磁通密度为1.78T,均满足样机要求。搭建电力变压器振动信息采集平台,对该变压器进行空载、负载及带载试验并进行数据分析。试验结果表明:负载条件下,油箱表面振动加速度频率主要集中在100Hz。空载条件下,频率主要集中在100Hz-500Hz。带载条件下,频率主要集中在100Hz-500Hz,验证了仿真模型的准确性。提出独立分量分析法,将空载和负载试验的振动信号进行随机混合再分离,分离后得到振动信号的时域及频谱分布与原信号基本一致,验证了该算法的有效性。接下来提出一种基于快速总体平均经验模态分解(FAST-Ensemble Empirical Mode Decomposition,FEEMD)和负熵准则的单通道盲源分离技术,将该方法应用于挂网运行变压器的振动信号分析,能够有效的辨识出绕组的振动信号,进而实现绕组振动信号的在线辨识。更多还原

【Abstract】 In the power system,the operational status of transformers significantly impacts the stable operation of the grid.During prolonged operation,transformers inevitably experience various disturbances,which can lead to faults;among these,winding faults constitute a substantial proportion.Although vibration analysis method can achieve online identification of transformer winding vibration signals,the exploration of the vibration characteristics of the equipment body and oil tank remains inadequate.Therefore,this thesis conducts an in-depth study on the vibration characteristics of a 220k V oil-immersed power transformer and the online identification of its windings,employing a combined approach of simulation analysis,experimental verification,and signal feature extraction.The generation mechanism and transmission process of vibrations in transformer cores and windings were systematically studied,with a particular focus on the impact of insulating oil on tank vibrations.The transmission of mechanical vibrations through the oil medium was equated to the transmission of sound waves,while the tank was simplified as an infinitely large single-plate structure.Take full account of the transmitted and reflected waves generated by the sound waves.Acoustic mathematical models for the transmission of vibration waves in insulating oil,as well as mathematical models for tank vibrations,were established.Using the finite element simulation method,a three-dimensional simulation model of the transformer was established.During the analysis process,by setting up the electric field,magnetic field,pressure acoustics,and solid mechanics modules,a multi-physics coupled simulation analysis of the transformer under different operating conditions was achieved.Specifically,the analysis focused on the magnetic field distribution patterns and vibration characteristics of the core and windings for two operating conditions:no-load and load.Based on this,the vibration distribution characteristics on the surface of the tank under no-load,load,and loaded conditions were calculated respectively.The results indicate that under rated load conditions,the current amplitude on the high-voltage side of the transformer is 683.8 A,with an error of 7.0%compared to the design value,while the current amplitude on the low-voltage side is 2083.3 A,with an error of 2.2%compared to the design value.Additionally,the winding leakage flux is measured at0.403 T.Under rated no-load conditions,the voltage amplitude on the high-voltage side of the transformer is 316.5 k V,with a deviation of 1.7%compared to the design value,while the voltage amplitude on the low-voltage side is 99.8 k V,with a deviation of 2.3%compared to the design value.The magnetic flux density in the core is 1.78 T,all of which meet the prototype requirements.This study constructs a vibration information acquisition platform for power transformers and conducts no-load,load,and on-load tests on the transformer,followed by data analysis.The test results indicate that under load conditions,the vibration acceleration frequency on the tank surface is primarily concentrated at 100 Hz.Under no-load conditions,the frequency is predominantly concentrated between 100 Hz and 500 Hz.Under on-load conditions,the frequency is mainly concentrated between 100 Hz and 500 Hz,thereby verifying the accuracy of the simulation model.Independent Component Analysis(ICA)method is proposed,which randomly mixes and subsequently separates the vibration signals from no-load and load tests.The time-domain and frequency spectrum distributions of the separated vibration signals are found to be consistent with the original signals,thereby verifying the effectiveness of the algorithm.Subsequently,a single-channel blind source separation technique based on Fast Ensemble Empirical Mode Decomposition(FEEMD)and the negentropy criterion is introduced.When applied to the analysis of vibration signals from grid-operated transformers,this algorithm effectively identifys the winding vibration signals,facilitating the online identification of these signals.更多还原