电力变压器振动信号分离方法研究

以独立成分分析(ICA)为代表的主流盲分离技术对信号独立性要求较高,难以分离具有高度相关性的变压器铁芯与绕组振动信号。为了分离变压器铁芯和绕组振动信号,建立了变压器振动信号混合模型,在该模型基础上提出了一种基于时频比盲源分离算法(TIFORM-BSS)的变压器振动信号分离方法。将该方法分别应用于分离人工混合后的110 k V三相变压器油箱壁信号和实际运行中的500 k V单相变压器油箱壁振动信号,结果表明该方法能够有效分离具有强相关性的变压器绕组和铁芯振动信号。

Research on vibration signal separation method of power transformer

In conventional Blind Source Separation methods, especially the Independent Component Analysis technique, the sources are supposed to be statistically independent with each other. Thus, the extraction of the highly correlated winding vibration and core vibration becomes complex and difficult. In order to extract the source vibrations generated by the core and winding, a mixing model based on single channel vibration is presented and a new fashioned method based on TIFROM (Time Frequency Ratio of Mixtures) method is proposed in this paper. The method was applied to separate vibration of an 110 kV three phase transformer and a running 500 kV single phase power transformer. Core and winding vibrations were successfully retrieved. The experimental results indicate that this approach has significant performance in blind separation of the highly correlated signals.