经验小波变换在变压器铁心松动故障诊断中的研究

Study on Fault Diagnosis of Transformer Core Looseness by Empirical Wavelet Transform

针对变压器铁心松动故障诊断中现有的信号处理方法存在模态混叠或受参数影响等问题,引入了经验小波变换方法,并提出了直接阈值边界划分方法对经验小波变换进行改进。对不同边界划分方法进行了对比研究,验证了直接阈值法可将信号的各个主要成分都能准确划分到各自分量之中的优点。以铁心松动故障为实例,通过计算改进型经验小波变换之后的能量熵来判别铁心的松动状态,发现随着铁心压紧力的减小,EWT能量熵明显增加,并且压紧力矩在20N·m左右时发生突变。用改进型经验小波变换的故障诊断方法可有效准确地判断铁心松动故障,对实际运行中变压器的故障监测有重要意义。

In the fault diagnosis of transformer core looseness, the conventional empirical wavelet transform （EWT） cannot implement the decomposition of some key frequency components. An improved empirical wavelet transform with a direct threshold segment method is introduced in this paper to solve the problem. Different segment methods of spectrum are studied, and the results verify the advantages of the direct threshold method, which can accurately divide the main ingredient of the signal into their respective components. Taking the core looseness fault as the example, the energy entropy is calculated after the empirical wavelet transform~ It is found that the EWT energy entropy increases with the decrease of core pressing force, and the mutation occurs around 20N ＂m. The fault diagnosis method based on improved empirical wavelet transform can be used to identify the fault of core looseness effectively and accurately, and it is important to the practical application of the transformer online monitoring.