Adaptive Separation of Vibration Signal of On-Load Tap Changer Based on Independent Component Analysis and Endpoint Detection

As an effective manner of blind source separation,independent component analysis (ICA) can rapidly and accurately implement the separation and recovery of source signals.In order to effectively separate the vibration signal of on-load tap changer (OLTC) of power transformer from the vibration signals of tank surface of power transformer,the second order blind identification in classical ICA algorithm is adopted to separate vibration signal of OLTC,and combining with endpoint detection algorithm of impact signal and phase difference detection algorithm of cross-correlation function the adaptive signal separation is implemented.Experimental results show that the proposed method can accurately and effectively extract OLTC vibration signal from tank surface vibration signals of power transformer.