摘要:AbstractDistribution system operators (DSOs) inspect electrical components in distribution systems and decide necessary actions for the inspection results to sustain the reliability of power supply. These results have been accumulated in databases as the inspection and the maintenance records. The authors propose to utilize the inspection and the maintenance records in estimation of operating lifetime of newly constructed electrical components in the distribution systems. A decision tree learning analyzes the relationship between attributes of the target components and the operating lifetime obtained by the inspection and the maintenance records, and as a result, constructs a tree-like classification or regression model. The resulting model can estimate the operating lifetime of newly constructed components using the initial information (the set of the attributes) only. Usefulness of the estimation method is verified on actual records stored in a Japanese electric power company. In addition, factors that have strong influence on the operating lifetime are specified by the feature of decision tree.
关键词:Keywordsdistribution equipmentlifetime estimationdecision treesinspection recordmaintenance recordbig data