Health monitoring of metallized film capacitors using Neo Fuzzy Neural approach

Recently, special attention has been devoted to the health monitoring aspect of power electronics devices since it became an essential step to ensure and guarantee the availability of the overall power electronics systems. Dc-link capacitors, despite their type or technology are one of the main devices to monitor since they are responsible of more than 30% of the power electronics equipment breakdowns. In this paper, a noninvasive method is proposed in order to detect the capacitors electrical parameter evolution using the Neo Fuzzy Neural approach. This latter is based on the real-time processing of the component's impedance and the ageing degradation data provided from accelerated ageing tests. A set of 10 polymer film capacitors was tested to validate the proposed approach.