Detection of Transformer Cores Looseness Using RQA

This paper proposes a recurrence qualification analysis (RQA) based method which is a useful tool to detect the subtle nonlinear changes in time series data to monitor the decreasing clamping forces of transformer iron cores. Firstly, suitable parameters including the distance threshold ¿ for RQA are chosen according to the dynamical structures of the core vibration signals, and then the signals sampled from transformer cores under different clamping conditions are analyzed using cross recurrence plot (CRP) and RQA. Finally, global and local RQA metrics for measuring the variation trend of the core clamping forces are computed from multivariate data and univariate data respectively. The vibration analysis of two different transformer cores shows that the proposed method can detect dropping of the clamping pressure exerted on the cores and reflect the damage evolution of the cores effectively.