Transformer Vibration Feature Extraction Method Based on Recursive Graph Quantitative Analysis

Transformer condition monitoring has always been a very important subject, and the proposed vibration-based monitoring method provides a new idea for it. In the limited vibration information, fully excavating the transformer state information contained in it is still an important task at present. This paper considers that the traditional frequency domain analysis method will ignore some frequency information of signals, and drawing on the experience of the construction field such as bridges, uses phase space reconstruction technology and quantitative recursive graph method based on the time series of vibration data to restore the original response characteristics of the vibration system. On this basis, using the principal component analysis method and comparing with the typical stationary signal quantifies the vibration stability of the transformer. Considering the actual operation of the transformer, the stability evaluation method proposed in this paper does reflect the stability of the transformer.