Avaliação da assertividade de projetistas em relação à previsão do impacto dos desligamentos

Abstract

The energy distributors’ concession contracts require, among other obligations, the maintenance of supply quality levels as well as their continuity. For this, companies need to constantly maintain and expand the system so that it meets the stipulated goals. In the state of Rio Grande do Sul, the electrical distribution network has approximately 40% of wooden poles and also several municipalities and a load center with only one power supply. There are also areas in constant economic expansion that have high energy demand. Given this scenario, distributors need to annually invest in the system through works. These, in turn, demand a large amount of material, in addition to scheduled shutdowns that hurt the same indicators that need to be improved. This difficult balance requires a good project and planning of works. This work will have its scope in the impact that the Project Technician has on the works and on the continuity indicators. In the design stage, it will be analyzed whether a particular work is viable from the point of view of quality and continuity indicators. Through the creation of a modeling tool for the project technician, the impact of scheduled shutdowns and the shutdown itself, in the execution of the work, will be foreseen. The human factor is currently decisive for the good planning of the works. In this way, this tool seeks to model the performance of the project technician and its influence on the correct planning of works. It is intended to assess whether the technicians are at a satisfactory nlevel according to their experience and difficulty of activities. The Project Technician’s modeling was done using the SVM and Neural Networks algorithms. Being the SVM the one that brought the best result with 54% of accuracy. This value represents a gain of 9% in relation to the current results. As this is a pioneering project, this work opens the way for further exploration of the topic, with the incorporation of new variables, it is estimated that an accuracy of up to 20%.