Optimization of maintenance management of trees in powerdistribution systems

We propose a data-analysis-based methodology for maintaining trees that affect power distribution networks. From the information captured in the field, we use fuzzy-logic-based techniques and models together with genetic algorithms so as to focus maintenance activities on the optimal time and place. The strategy is aimed at two key aspects: first, the history of failure events caused by trees on medium-voltage power supply systems; this information is used to calculate quality indicators such as energy-not-supplied (ENS) and number-of-users (NU) affected due to falling trees only. In this case we use a fuzzy-logic-based system to weight and find the critical circuits on which forest maintenance should be performed. Second, we construct an objective function based on tree-to-network distances and growth patterns of trees (classified per species) in order to simulate forest maintenance; this is achieved using a genetic algorithm to determine optimum pruning cycles for a given power supply system. We use real historical data of medium voltage circuits for system setting as well as for performance evaluation.