Research On Transformer On-Load Tap Changer's Vibration Signal Separation

Blind source separation is an approach to recover unobserved signals or "sources" from observed mixtures exploiting only the assumption of mutual independence between the signals. It has been applied to various areas, e.g., speech enhancement with multiple microphones, biological signal separation, such as EEG, MEG and others, image processing. This thesis describes in detail the basic theory of blind source separation, then studies blind separation of linearly mixed sources and convolved signals. Finally, applies the SOBI algorithm to the transformer's signal separation. Main works can be summarized as follows:In this paper, the ICA of linearly instantaneous mixed sources is discussed in simulation. The simulation compared the separation performance,range of application and iteration speed pf all algorithm.We also research the convolutive blind separation. Especially we discussed a frequency domain algorithm which used second-order statistics as principle. The simulation results show a good separation quality.We applied the SOBI algorithm to separating the On-Load Tap Changer's vibration signal from tank vibration signals of power transformers adaptively by combining SOBI algorithm and endpoint detection algorithm. The results of simulations and experiments show that the method can extract the OLTC vibration signal exactly and effectively.