Condition-Based Modelling of Events in the Life Cycle of Medium-Voltage Substations for Optimised Asset Management Strategies

Inhalt:
In this work, event specific condition indices are developed for relevant events in the life cycle of medium-voltage substations (MVSS), which consider the results of the regularly conducted inspections of distribution system operators (DSOs) with regard to a possible event occurrence. The events under consideration have already been defined in cooperation with several DSOs in the context of value-based asset management (AM) for MVSS in [1]. To derive probabilities of occurrence in the form of hazard rates for the events, the event specific condition indices are used as covariates for a survival regression (SR). The SR is applied to the data sets of two DSOs using both Cox proportional hazards (CPH) and a Weibull distribution based accelerated failure times (AFT) model. SR as part of the survival analysis (SA) is commonly used for medical studies to predict the expected lifetime of patient groups considering several parameters and to compare different treatment methods [2]. In this work the SR is applied to real failure data for equipment in the distribution system and to model the relevant events. The results show that the SR can be applied well to the data and that there is a significant correlation between the event specific condition of the equipment and the hazard rate for the events observed, both in the CPH and in the AFT model. In the further progress of the research project the hazard rates will be used to calculate an expected risk value to be able to optimise AM strategies for MVSS with a reinforcement learning approach [3].