State-space analysis of partial discharge process

Abstract

We analyse the dynamics of the electric field and local surface charge for three kinds of partial discharge (PD) sources: point to dielectric gap, electrode bound cavity and electrical treeing. The data used in our analysis were simulated by a deterministic model proposed by Patsch and Hoof (1998). We demonstrate a dynamic system approach to the analysis of partial discharge data. In particular, we use time-delayed reconstruction of dynamics from a single observable scalar value. We calculate the dominant Lyapunov exponent for the three cases of PD sources. We conclude that the positive Lyapunov exponents obtained in our investigation indicate the possibility of chaotic behaviour. Finally, we address the essential problem of the state-space analysis from experimental PD data: Can the states be identified from the PD interval data? We demonstrate that the phase-space trajectory reconstructed from PD interval data (inter-spike intervals) reveals the deterministic character of the process and potentially can be used in PD identification.