The here reported work is part of a project on supporting grid resilience by asset management techniques. The present work focuses on support of decision-making after a few failures occurred that may be the start of many more. Methods are reviewed and new algorithms developed where the present IEEE/IEC standard does not provide. Two cases of early failures and wear-out are analyzed as examples for the data analytics.

This paper discusses a method for data analysis of early failures that are typically due to defects from production and/or installation. The approach is based on Weighted Linear Regression and aims at estimating the next failure moment and the failure probability in an interval. As an additional goal, methods are intended to be suitable for light computing devices.