Predictive maintenance attempts to evaluate the condition of equipment and predict the future trend of the equipment's aging, in order to reduce costs when compared to the two traditional approaches: corrective and preventive maintenance. This prediction requires an accurate prognostic model of aging. In solid insulation, the ultimate goal of prognostics is to predict the advent of failure, i.e., insulation breakdown, in terms of remaining useful life (RUL). One fault is electrical treeing, which is progressive thus leading to potentially catastrophic failure. Research has shown that diagnosis of faults can be achieved based on partial discharge (PD) monitoring , i.e., phase-resolved and pulse sequence analysis (PSA). This work will explore the extension of this concept towards predicting evolution of the defect: moving beyond diagnostics towards prognostics. To do this, there is a need for further investigation of prognostic features within PD characteristics leading up to breakdown. In this work, a needle-plane test arrangement was set up using a hypodermic needle and pre-formed silicone rubber as test samples. The visual observations and tree growth measurements were made using a digital microscope. PD data was captured using a radio frequency (RF) sensor and analysed using PSA. The main idea of the PSA approach is the strong relationship between two consecutive pulses caused by PD activities, which can give an understanding of the local degradation processes . As for electrical treeing, a breakdown indicator in PSA is the appearance of heavily clustered data points that lie diagonally in scatter plots of the differential ratio of voltage and time of consecutive charges (Un = Δun/Δtn) [2,3]. Figure 1 shows an example of a plot that changed to a diagonal line after 14 hours of aging time. This paper investigates the formation of the diagonal line based on the distribution of the plot from the start of electrical treeing until breakdown occurs. Finally, statistical features of the PSA plot are given and will be used for lifetime prediction of insulation samples in future work.
|Title of host publication||Universities High Voltage Network (UHVnet) Colloquium|
|Place of Publication||Glasgow|
|Publication status||Published - Jan 2014|
- prognostic indicators
- electrical treeing
- solid insulation
- pulse sequence analysis